Libguestfs - Mar 2018 - Re: [PATCH v6] v2v: Add -o rhv-upload output mode.

v5 was here:
https://www.redhat.com/archives/libguestfs/2018-March/msg00032.html

There is only a single patch in this version because the other
patches went upstream.

This patch adds the virt-v2v -o rhv-upload mode
(https://bugzilla.redhat.com/show_bug.cgi?id=1557273).

Compared to v5, this adds the ability to make zero, trim and flush
requests to the oVirt imageio server
(https://www.redhat.com/archives/libguestfs/2018-March/msg00119.html).
However this is untested, although I believe it matches the
specification (https://gerrit.ovirt.org/#/c/89022/).  As a fallback it
converts some zero requests into pwrites and ignores others, hopefully
when it is safe to do so, and it doesn't make trim or flush requests
at all.

Also this adds the capability of setting the cluster.

However it doesn't allow you to choose the datacenter yet.  Also -of
qcow2 probably doesn't work and is definitely not tested.

Rich.

PROBLEMS:
 - -of qcow2 does not work, with multiple problems
    * needs to set NBD size to something larger than virtual size

This adds a new output mode to virt-v2v.  virt-v2v -o rhv-upload
streams images directly to an oVirt or RHV >= 4 Data Domain using the
oVirt SDK v4.  It is more efficient than -o rhv because it does not
need to go via the Export Storage Domain, and is possible for humans
to use unlike -o vdsm.

The implementation uses the Python SDK (‘ovirtsdk4’ module).  An
nbdkit Python 3 plugin translates NBD calls from qemu into HTTPS
requests to oVirt via the SDK.
---
 .gitignore                                |   3 +
 v2v/Makefile.am                           |  28 ++-
 v2v/cmdline.ml                            |  46 ++++
 v2v/embed.sh                              |  45 ++++
 v2v/output_rhv_upload.ml                  | 339 +++++++++++++++++++++++++
 v2v/output_rhv_upload.mli                 |  27 ++
 v2v/output_rhv_upload_createvm_source.mli |  19 ++
 v2v/output_rhv_upload_plugin_source.mli   |  19 ++
 v2v/output_rhv_upload_precheck_source.mli |  19 ++
 v2v/rhv-upload-createvm.py                |  85 +++++++
 v2v/rhv-upload-plugin.py                  | 394 ++++++++++++++++++++++++++++++
 v2v/rhv-upload-precheck.py                |  72 ++++++
 v2v/test-v2v-python-syntax.sh             |  27 ++
 v2v/virt-v2v.pod                          | 125 ++++++++--
 14 files changed, 1233 insertions(+), 15 deletions(-)

diff --git a/.gitignore b/.gitignore
index d72447d1d..211376eef 100644
--- a/.gitignore
+++ b/.gitignore
@@ -654,6 +654,9 @@ Makefile.in
 /utils/qemu-speed-test/qemu-speed-test
 /v2v/.depend
 /v2v/oUnit-*
+/v2v/output_rhv_upload_createvm_source.ml
+/v2v/output_rhv_upload_plugin_source.ml
+/v2v/output_rhv_upload_precheck_source.ml
 /v2v/real-*.d/
 /v2v/real-*.img
 /v2v/real-*.xml
diff --git a/v2v/Makefile.am b/v2v/Makefile.am
index c2eb31097..987ad8319 100644
--- a/v2v/Makefile.am
+++ b/v2v/Makefile.am
@@ -22,12 +22,19 @@ generator_built = \
 	uefi.mli
 
 BUILT_SOURCES = \
-	$(generator_built)
+	$(generator_built) \
+	output_rhv_upload_createvm_source.ml \
+	output_rhv_upload_plugin_source.ml \
+	output_rhv_upload_precheck_source.ml
 
 EXTRA_DIST = \
 	$(SOURCES_MLI) $(SOURCES_ML) $(SOURCES_C) \
 	copy_to_local.ml \
 	copy_to_local.mli \
+	embed-code.sh \
+	rhv-upload-createvm.py \
+	rhv-upload-plugin.py \
+	rhv-upload-precheck.py \
 	v2v_slow_unit_tests.ml \
 	v2v-slow-unit-tests.sh \
 	v2v_unit_tests.ml \
@@ -64,6 +71,10 @@ SOURCES_MLI = \
 	output_null.mli \
 	output_qemu.mli \
 	output_rhv.mli \
+	output_rhv_upload.mli \
+	output_rhv_upload_createvm_source.mli \
+	output_rhv_upload_plugin_source.mli \
+	output_rhv_upload_precheck_source.mli \
 	output_vdsm.mli \
 	parse_ovf_from_ova.mli \
 	parse_libvirt_xml.mli \
@@ -116,6 +127,10 @@ SOURCES_ML = \
 	output_local.ml \
 	output_qemu.ml \
 	output_rhv.ml \
+	output_rhv_upload_createvm_source.ml \
+	output_rhv_upload_plugin_source.ml \
+	output_rhv_upload_precheck_source.ml \
+	output_rhv_upload.ml \
 	output_vdsm.ml \
 	inspect_source.ml \
 	target_bus_assignment.ml \
@@ -126,6 +141,15 @@ SOURCES_C = \
 	libvirt_utils-c.c \
 	qemuopts-c.c
 
+# These files are generated and contain rhv-upload-*.py embedded as an
+# OCaml string.
+output_rhv_upload_createvm_source.ml: rhv-upload-createvm.py
+	./embed.sh code $^ $@
+output_rhv_upload_plugin_source.ml: rhv-upload-plugin.py
+	./embed.sh code $^ $@
+output_rhv_upload_precheck_source.ml: rhv-upload-precheck.py
+	./embed.sh code $^ $@
+
 if HAVE_OCAML
 
 bin_PROGRAMS = virt-v2v virt-v2v-copy-to-local
@@ -295,6 +319,7 @@ TESTS_ENVIRONMENT = $(top_builddir)/run --test
 
 TESTS = \
 	test-v2v-docs.sh \
+	test-v2v-python-syntax.sh \
 	test-v2v-i-ova-bad-sha1.sh \
 	test-v2v-i-ova-bad-sha256.sh \
 	test-v2v-i-ova-formats.sh \
@@ -490,6 +515,7 @@ EXTRA_DIST += \
 	test-v2v-print-source.expected \
 	test-v2v-print-source.sh \
 	test-v2v-print-source.xml \
+	test-v2v-python-syntax.sh \
 	test-v2v-conversion-of.sh \
 	test-v2v-sound.sh \
 	test-v2v-sound.xml \
diff --git a/v2v/cmdline.ml b/v2v/cmdline.ml
index 14c5c0dca..1b3c10be8 100644
--- a/v2v/cmdline.ml
+++ b/v2v/cmdline.ml
@@ -67,6 +67,10 @@ let parse_cmdline ()    let output_password = ref None in
   let output_storage = ref None in
   let password_file = ref None in
+  let rhv_cafile = ref None in
+  let rhv_cluster = ref None in
+  let rhv_direct = ref false in
+  let rhv_verifypeer = ref false in
   let vddk_config = ref None in
   let vddk_cookie = ref None in
   let vddk_libdir = ref None in
@@ -145,6 +149,8 @@ let parse_cmdline ()      | "disk" |
"local" -> output_mode := `Local
     | "null" -> output_mode := `Null
     | "ovirt" | "rhv" | "rhev" -> output_mode
:= `RHV
+    | "ovirt-upload" | "ovirt_upload" |
"rhv-upload" | "rhv_upload" ->
+       output_mode := `RHV_Upload
     | "qemu" -> output_mode := `QEmu
     | "vdsm" -> output_mode := `VDSM
     | s ->
@@ -233,6 +239,12 @@ let parse_cmdline ()      [ L"print-target" ],
Getopt.Set print_target,
                                     s_"Print target and stop";
     [ L"qemu-boot" ], Getopt.Set qemu_boot, s_"Boot in qemu (-o
qemu only)";
+    [ L"rhv-cafile" ], Getopt.String ("ca.pem",
set_string_option_once "--rhv-cafile" rhv_cafile),
+                                    s_"For -o rhv-upload, set ‘ca.pem’
file";
+    [ L"rhv-cluster" ], Getopt.String ("clustername",
set_string_option_once "--rhv-cluster" rhv_cluster),
+                                    s_"For -o rhv-upload, set cluster
name";
+    [ L"rhv-direct" ], Getopt.Set rhv_direct, s_"Use direct
transfer mode";
+    [ L"rhv-verifypeer" ], Getopt.Set rhv_verifypeer, s_"Verify
server identity";
     [ L"root" ],     Getopt.String ("ask|... ",
set_root_choice),
                                     s_"How to choose root
filesystem";
     [ L"vddk-config" ], Getopt.String ("filename",
set_string_option_once "--vddk-config" vddk_config),
@@ -327,6 +339,10 @@ read the man page virt-v2v(1).
   let print_source = !print_source in
   let print_target = !print_target in
   let qemu_boot = !qemu_boot in
+  let rhv_cafile = !rhv_cafile in
+  let rhv_cluster = !rhv_cluster in
+  let rhv_direct = !rhv_direct in
+  let rhv_verifypeer = !rhv_verifypeer in
   let root_choice = !root_choice in
   let vddk_options        { vddk_config = !vddk_config;
@@ -557,6 +573,36 @@ read the man page virt-v2v(1).
       Output_rhv.output_rhv os output_alloc,
       output_format, output_alloc
 
+    | `RHV_Upload ->
+      let output_conn +        match output_conn with
+        | None ->
+           error (f_"-o rhv-upload: use ‘-oc’ to point to the oVirt or RHV
server REST API URL, which is usually https://servername/ovirt-engine/api")
+        | Some oc -> oc in
+      (* In theory we could make the password optional in future. *)
+      let output_password +        match output_password with
+        | None ->
+           error (f_"-o rhv-upload: output password file was not
specified, use ‘-op’ to point to a file which contains the password used to
connect to the oVirt or RHV server")
+        | Some op -> op in
+      let os +        match output_storage with
+        | None ->
+           error (f_"-o rhv-upload: output storage was not specified, use
‘-os’");
+        | Some os -> os in
+      if qemu_boot then
+        error_option_cannot_be_used_in_output_mode "rhv-upload"
"--qemu-boot";
+      let rhv_cafile +        match rhv_cafile with
+        | None ->
+           error (f_"-o rhv-upload: must use ‘--rhv-cafile’ to supply the
path to the oVirt or RHV server’s ‘ca.pem’ file")
+        | Some rhv_cafile -> rhv_cafile in
+      Output_rhv_upload.output_rhv_upload output_alloc output_conn
+                                          output_password os
+                                          rhv_cafile rhv_cluster
+                                          rhv_direct rhv_verifypeer,
+      output_format, output_alloc
+
     | `VDSM ->
       if output_password <> None then
         error_option_cannot_be_used_in_output_mode "vdsm"
"-op";
diff --git a/v2v/embed.sh b/v2v/embed.sh
new file mode 100755
index 000000000..0a65cd428
--- /dev/null
+++ b/v2v/embed.sh
@@ -0,0 +1,45 @@
+#!/bin/bash -
+# Embed code or other content into an OCaml file.
+# Copyright (C) 2018 Red Hat Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+# Embed code or other content into an OCaml file.
+#
+# It is embedded into a string.  As OCaml string literals have virtually
+# no restrictions on length or content we only have to escape double
+# quotes for backslash characters.
+
+if [ $# -ne 3 ]; then
+    echo "embed.sh identifier input output"
+    exit 1
+fi
+
+ident="$1"
+input="$2"
+output="$3"
+
+rm -f "$output" "$output"-t
+
+exec >"$output"-t
+
+echo "(* Generated by embed.sh from $input *)"
+echo
+echo let "$ident" = '"'
+sed -e 's/\(["\]\)/\\\1/g' < "$input"
+echo '"'
+
+chmod -w "$output"-t
+mv "$output"-t "$output"
diff --git a/v2v/output_rhv_upload.ml b/v2v/output_rhv_upload.ml
new file mode 100644
index 000000000..77f0afa40
--- /dev/null
+++ b/v2v/output_rhv_upload.ml
@@ -0,0 +1,339 @@
+(* virt-v2v
+ * Copyright (C) 2009-2018 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *)
+
+open Printf
+open Unix
+
+open Std_utils
+open Tools_utils
+open Unix_utils
+open Common_gettext.Gettext
+
+open Types
+open Utils
+
+let python3 = "python3" (* Defined by PEP 394 *)
+let pidfile_timeout = 30
+let finalization_timeout = 5*60
+
+class output_rhv_upload output_alloc output_conn
+                        output_password output_storage
+                        rhv_cafile rhv_cluster
+                        rhv_direct rhv_verifypeer +  (* Create a temporary
directory which will be deleted on exit. *)
+  let tmpdir +    let base_dir = (open_guestfs ())#get_cachedir () in
+    let t = Mkdtemp.temp_dir ~base_dir "rhvupload." in
+    rmdir_on_exit t;
+    t in
+
+  let diskid_file_of_id id = tmpdir // sprintf "diskid.%d" id in
+
+  (* Write the Python precheck, plugin and create VM to a temporary file. *)
+  let precheck +    let precheck = tmpdir // "rhv-upload-precheck.py"
in
+    with_open_out
+      precheck
+      (fun chan -> output_string chan
Output_rhv_upload_precheck_source.code);
+    precheck in
+  let plugin +    let plugin = tmpdir // "rhv-upload-plugin.py" in
+    with_open_out
+      plugin
+      (fun chan -> output_string chan Output_rhv_upload_plugin_source.code);
+    plugin in
+  let createvm +    let createvm = tmpdir // "rhv-upload-createvm.py"
in
+    with_open_out
+      createvm
+      (fun chan -> output_string chan
Output_rhv_upload_createvm_source.code);
+    createvm in
+
+  (* Is SELinux enabled and enforcing on the host? *)
+  let have_selinux +    0 = Sys.command "getenforce 2>/dev/null | grep
-isq Enforcing" in
+
+  (* Check that nbdkit is available and new enough. *)
+  let error_unless_nbdkit_working () +    if 0 <> Sys.command
"nbdkit --version >/dev/null" then
+      error (f_"nbdkit is not installed or not working.  It is required to
use ‘-o rhv-upload’.  See \"OUTPUT TO RHV\" in the virt-v2v(1)
manual.");
+
+    (* Check it's a new enough version.  The latest features we
+     * require are ‘--exit-with-parent’ and ‘--selinux-label’, both
+     * added in 1.1.14.  (We use 1.1.16 as the minimum here because
+     * it also adds the selinux=yes|no flag in --dump-config).
+     *)
+    let lines = external_command "nbdkit --help" in
+    let lines = String.concat " " lines in
+    if String.find lines "exit-with-parent" == -1 ||
+       String.find lines "selinux-label" == -1 then
+      error (f_"nbdkit is not new enough, you need to upgrade to nbdkit ≥
1.1.16")
+  in
+
+  (* Check that the python3 plugin is installed and working
+   * and can load the plugin script.
+   *)
+  let error_unless_nbdkit_python3_working () +    let cmd = sprintf
"nbdkit %s %s --dump-plugin >/dev/null"
+                      python3 (quote plugin) in
+    if Sys.command cmd <> 0 then
+      error (f_"nbdkit Python 3 plugin is not installed or not working. 
It is required if you want to use ‘-o rhv-upload’.
+
+See also \"OUTPUT TO RHV\" in the virt-v2v(1) manual.")
+  in
+
+  (* Check that nbdkit was compiled with SELinux support (for the
+   * --selinux-label option).
+   *)
+  let error_unless_nbdkit_compiled_with_selinux () +    let lines =
external_command "nbdkit --dump-config" in
+    (* In nbdkit <= 1.1.15 the selinux attribute was not present
+     * at all in --dump-config output so there was no way to tell.
+     * Ignore this case because there will be an error later when
+     * we try to use the --selinux-label parameter.
+     *)
+    if List.mem "selinux=no" (List.map String.trim lines) then
+      error (f_"nbdkit was compiled without SELinux support.  You will
have to recompile nbdkit with libselinux-devel installed, or else set SELinux to
Permissive mode while doing the conversion.")
+  in
+
+  (* JSON parameters which are invariant between disks. *)
+  let json_params = [
+    "output_conn", JSON.String output_conn;
+    "output_password", JSON.String output_password;
+    "output_storage", JSON.String output_storage;
+    "output_sparse", JSON.Bool (match output_alloc with
+                                | Sparse -> true
+                                | Preallocated -> false);
+    "rhv_cafile", JSON.String rhv_cafile;
+    "rhv_cluster", JSON.String (Option.default "Default"
rhv_cluster);
+    "rhv_direct", JSON.Bool rhv_direct;
+
+    (* The 'Insecure' flag seems to be a number with various possible
+     * meanings, however we just set it to True/False.
+     *
+     *
https://github.com/oVirt/ovirt-engine-sdk/blob/19aa7070b80e60a4cfd910448287aecf9083acbe/sdk/lib/ovirtsdk4/__init__.py#L395
+     *)
+    "insecure", JSON.Bool (not rhv_verifypeer);
+  ] in
+
+  (* nbdkit command line args which are invariant between disks. *)
+  let nbdkit_args +    let args = [
+      "nbdkit";
+
+      "--foreground";           (* run in foreground *)
+      "--exit-with-parent";     (* exit when virt-v2v exits *)
+      "--newstyle";             (* use newstyle NBD protocol *)
+      "--exportname"; "/";
+
+      "python3";                (* use the nbdkit Python 3 plugin *)
+      plugin;                   (* Python plugin script *)
+    ] in
+    let args = if verbose () then args @ ["--verbose"] else args in
+    let args +      (* label the socket so qemu can open it *)
+      if have_selinux then
+        args @ ["--selinux-label";
"system_u:object_r:svirt_t:s0"]
+      else args in
+    args in
+
+object
+  inherit output
+
+  method precheck () +    error_unless_nbdkit_working ();
+    error_unless_nbdkit_python3_working ();
+    if have_selinux then
+      error_unless_nbdkit_compiled_with_selinux ()
+
+  method as_options +    "-o rhv-upload" ^
+    (match output_alloc with
+     | Sparse -> "" (* default, don't need to print it *)
+     | Preallocated -> " -oa preallocated") ^
+    sprintf " -oc %s -op %s -os %s"
+            output_conn output_password output_storage
+
+  method supported_firmware = [ TargetBIOS ]
+
+  method prepare_targets source targets +    let output_name = source.s_name in
+    let json_params +      ("output_name", JSON.String output_name)
:: json_params in
+
+    (* Python code prechecks.  These can't run in #precheck because
+     * we need to know the name of the virtual machine.
+     *)
+    let json_param_file = tmpdir // "params.json" in
+    with_open_out
+      json_param_file
+      (fun chan -> output_string chan (JSON.string_of_doc json_params));
+    if run_command [ python3; precheck; json_param_file ] <> 0 then
+      error (f_"failed server prechecks, see earlier errors");
+
+    (* Create an nbdkit instance for each disk and set the
+     * target URI to point to the NBD socket.
+     *)
+    List.map (
+      fun t ->
+        let id = t.target_overlay.ov_source.s_disk_id in
+        let disk_name = sprintf "%s-%03d" output_name id in
+        let json_params +          ("disk_name", JSON.String
disk_name) :: json_params in
+
+        let disk_format +          match t.target_format with
+          | ("raw" | "qcow2") as fmt -> fmt
+          | _ ->
+             error (f_"rhv-upload: -of %s: Only output format ‘raw’ or
‘qcow2’ is supported.  If the input is in a different format then force one of
these output formats by adding either ‘-of raw’ or ‘-of qcow2’ on the command
line.")
+                   t.target_format in
+        let json_params +          ("disk_format", JSON.String
disk_format) :: json_params in
+
+        let disk_size = t.target_overlay.ov_virtual_size in
+        let json_params +          ("disk_size", JSON.Int64
disk_size) :: json_params in
+
+        (* Ask the plugin to write the disk ID to a special file. *)
+        let diskid_file = diskid_file_of_id id in
+        let json_params +          ("diskid_file", JSON.String
diskid_file) :: json_params in
+
+        (* Write the JSON parameters to a file. *)
+        let json_param_file = tmpdir // sprintf "params%d.json" id in
+        with_open_out
+          json_param_file
+          (fun chan -> output_string chan (JSON.string_of_doc json_params));
+
+        let sock = tmpdir // sprintf "nbdkit%d.sock" id in
+        let pidfile = tmpdir // sprintf "nbdkit%d.pid" id in
+
+        (* Add common arguments to per-target arguments. *)
+        let args +          nbdkit_args @ [ "--pidfile"; pidfile;
+                          "--unix"; sock;
+                          sprintf "params=%s" json_param_file ] in
+
+        (* Print the full command we are about to run when debugging. *)
+        if verbose () then (
+          eprintf "running nbdkit:\n";
+          List.iter (fun arg -> eprintf " %s" (quote arg)) args;
+          prerr_newline ()
+        );
+
+        (* Start an nbdkit instance in the background.  By using
+         * --exit-with-parent we don't have to worry about clean-up.
+         *)
+        let args = Array.of_list args in
+        let pid = fork () in
+        if pid = 0 then (
+          (* Child process (nbdkit). *)
+          execvp "nbdkit" args
+        );
+
+        (* Wait for the pidfile to appear so we know that nbdkit
+         * is listening for requests.
+         *)
+        if not (wait_for_file pidfile pidfile_timeout) then (
+          if verbose () then
+            error (f_"nbdkit did not start up.  See previous debugging
messages for problems.")
+          else
+            error (f_"nbdkit did not start up.  There may be errors
printed by nbdkit above.
+
+If the messages above are not sufficient to diagnose the problem then add the
‘virt-v2v -v -x’ options and examine the debugging output carefully.")
+        );
+
+        if have_selinux then (
+          (* Note that Unix domain sockets have both a file label and
+           * a socket/process label.  Using --selinux-label above
+           * only set the socket label, but we must also set the file
+           * label.
+           *)
+          ignore (
+              run_command ["chcon";
"system_u:object_r:svirt_image_t:s0";
+                           sock]
+          );
+        );
+        (* ... and the regular Unix permissions, in case qemu is
+         * running as another user.
+         *)
+        chmod sock 0o777;
+
+        (* Tell ‘qemu-img convert’ to write to the nbd socket which is
+         * connected to nbdkit.
+         *)
+        let json_params = [
+          "file.driver", JSON.String "nbd";
+          "file.path", JSON.String sock;
+          "file.export", JSON.String "/";
+        ] in
+        let target_file +          TargetURI ("json:" ^
JSON.string_of_doc json_params) in
+        { t with target_file }
+    ) targets
+
+  method create_metadata source targets _ guestcaps inspect target_firmware +  
(* Get the UUIDs of each disk image.  These files are written
+     * out by the nbdkit plugins on successful finalization of the
+     * transfer.
+     *)
+    let nr_disks = List.length targets in
+    let image_uuids +      List.map (
+        fun t ->
+          let id = t.target_overlay.ov_source.s_disk_id in
+          let diskid_file = diskid_file_of_id id in
+          if not (wait_for_file diskid_file finalization_timeout) then
+            error (f_"transfer of disk %d/%d failed, see earlier error
messages")
+                  (id+1) nr_disks;
+          let diskid = read_whole_file diskid_file in
+          diskid
+      ) targets in
+
+    (* We don't have the storage domain UUID, but instead we write
+     * in a magic value which the Python code (which can get it)
+     * will substitute.
+     *)
+    let sd_uuid = "@SD_UUID@" in
+
+    (* The volume and VM UUIDs are made up. *)
+    let vol_uuids = List.map (fun _ -> uuidgen ()) targets
+    and vm_uuid = uuidgen () in
+
+    (* Create the metadata. *)
+    let ovf +      Create_ovf.create_ovf source targets guestcaps inspect
+                            output_alloc
+                            sd_uuid image_uuids vol_uuids vm_uuid
+                            OVirt in
+    let ovf = DOM.doc_to_string ovf in
+
+    let json_param_file = tmpdir // "params.json" in
+    with_open_out
+      json_param_file
+      (fun chan -> output_string chan (JSON.string_of_doc json_params));
+
+    let ovf_file = tmpdir // "vm.ovf" in
+    with_open_out ovf_file (fun chan -> output_string chan ovf);
+    if run_command [ python3; createvm; json_param_file; ovf_file ] <> 0
then
+      error (f_"failed to create virtual machine, see earlier
errors")
+
+end
+
+let output_rhv_upload = new output_rhv_upload
+let () = Modules_list.register_output_module "rhv-upload"
diff --git a/v2v/output_rhv_upload.mli b/v2v/output_rhv_upload.mli
new file mode 100644
index 000000000..c35d83f7e
--- /dev/null
+++ b/v2v/output_rhv_upload.mli
@@ -0,0 +1,27 @@
+(* virt-v2v
+ * Copyright (C) 2009-2018 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *)
+
+(** [-o rhv-upload] target. *)
+
+val output_rhv_upload : Types.output_allocation -> string -> string ->
+                        string -> string -> string option -> bool
-> bool ->
+                        Types.output
+(** [output_rhv_upload output_alloc output_conn output_password output_storage
+     rhv_cafile rhv_cluster rhv_direct rhv_verifypeer]
+    creates and returns a new {!Types.output} object specialized for writing
+    output to oVirt or RHV directly via RHV APIs. *)
diff --git a/v2v/output_rhv_upload_createvm_source.mli
b/v2v/output_rhv_upload_createvm_source.mli
new file mode 100644
index 000000000..c1bafa15b
--- /dev/null
+++ b/v2v/output_rhv_upload_createvm_source.mli
@@ -0,0 +1,19 @@
+(* virt-v2v
+ * Copyright (C) 2018 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *)
+
+val code : string
diff --git a/v2v/output_rhv_upload_plugin_source.mli
b/v2v/output_rhv_upload_plugin_source.mli
new file mode 100644
index 000000000..c1bafa15b
--- /dev/null
+++ b/v2v/output_rhv_upload_plugin_source.mli
@@ -0,0 +1,19 @@
+(* virt-v2v
+ * Copyright (C) 2018 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *)
+
+val code : string
diff --git a/v2v/output_rhv_upload_precheck_source.mli
b/v2v/output_rhv_upload_precheck_source.mli
new file mode 100644
index 000000000..c1bafa15b
--- /dev/null
+++ b/v2v/output_rhv_upload_precheck_source.mli
@@ -0,0 +1,19 @@
+(* virt-v2v
+ * Copyright (C) 2018 Red Hat Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License as published by
+ * the Free Software Foundation; either version 2 of the License, or
+ * (at your option) any later version.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program; if not, write to the Free Software Foundation, Inc.,
+ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+ *)
+
+val code : string
diff --git a/v2v/rhv-upload-createvm.py b/v2v/rhv-upload-createvm.py
new file mode 100644
index 000000000..7a80fa5bd
--- /dev/null
+++ b/v2v/rhv-upload-createvm.py
@@ -0,0 +1,85 @@
+# -*- python -*-
+# oVirt or RHV upload create VM used by ‘virt-v2v -o rhv-upload’
+# Copyright (C) 2018 Red Hat Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License along
+# with this program; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+import json
+import logging
+import ovirtsdk4 as sdk
+import ovirtsdk4.types as types
+import sys
+import time
+
+from http.client import HTTPSConnection
+from urllib.parse import urlparse
+
+# Parameters are passed in via a JSON doc from the OCaml code.
+# Because this Python code ships embedded inside virt-v2v there
+# is no formal API here.
+params = None
+ovf = None                      # OVF file
+
+if len(sys.argv) != 3:
+    raise RuntimeError("incorrect number of parameters")
+
+# Parameters are passed in via a JSON document.
+with open(sys.argv[1], 'r') as fp:
+    params = json.load(fp)
+
+# What is passed in is a password file, read the actual password.
+with open(params['output_password'], 'r') as fp:
+    output_password = fp.read()
+output_password = output_password.rstrip()
+
+# Read the OVF document.
+with open(sys.argv[2], 'r') as fp:
+    ovf = fp.read()
+
+# Parse out the username from the output_conn URL.
+parsed = urlparse(params['output_conn'])
+username = parsed.username or "admin@internal"
+
+# Connect to the server.
+connection = sdk.Connection(
+    url = params['output_conn'],
+    username = username,
+    password = output_password,
+    ca_file = params['rhv_cafile'],
+    log = logging.getLogger(),
+    insecure = params['insecure'],
+)
+
+system_service = connection.system_service()
+
+# Get the storage domain UUID and substitute it into the OVF doc.
+sds_service = system_service.storage_domains_service()
+sd = sds_service.list(search=("name=%s" %
params['output_storage']))[0]
+sd_uuid = sd.id
+
+ovf.replace("@SD_UUID@", sd_uuid)
+
+vms_service = system_service.vms_service()
+vm = vms_service.add(
+    types.Vm(
+        cluster=types.Cluster(name = params['rhv_cluster']),
+        initialization=types.Initialization(
+            configuration = types.Configuration(
+                type = types.ConfigurationType.OVA,
+                data = ovf,
+            )
+        )
+    )
+)
diff --git a/v2v/rhv-upload-plugin.py b/v2v/rhv-upload-plugin.py
new file mode 100644
index 000000000..0ece05b94
--- /dev/null
+++ b/v2v/rhv-upload-plugin.py
@@ -0,0 +1,394 @@
+# -*- python -*-
+# oVirt or RHV upload nbdkit plugin used by ‘virt-v2v -o rhv-upload’
+# Copyright (C) 2018 Red Hat Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License along
+# with this program; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+import builtins
+import json
+import logging
+import ovirtsdk4 as sdk
+import ovirtsdk4.types as types
+import ssl
+import sys
+import time
+
+from http.client import HTTPSConnection
+from urllib.parse import urlparse
+
+# Timeout to wait for oVirt disks to change status, or the transfer
+# object to finish initializing [seconds].
+timeout = 5*60
+
+# Parameters are passed in via a JSON doc from the OCaml code.
+# Because this Python code ships embedded inside virt-v2v there
+# is no formal API here.
+params = None
+
+def config(key, value):
+    global params
+
+    if key == "params":
+        with builtins.open(value, 'r') as fp:
+            params = json.load(fp)
+    else:
+        raise RuntimeError("unknown configuration key '%s'" %
key)
+
+def config_complete():
+    if params is None:
+        raise RuntimeError("missing configuration parameters")
+
+def open(readonly):
+    # Parse out the username from the output_conn URL.
+    parsed = urlparse(params['output_conn'])
+    username = parsed.username or "admin@internal"
+
+    # Read the password from file.
+    with builtins.open(params['output_password'], 'r') as fp:
+        password = fp.read()
+    password = password.rstrip()
+
+    # Connect to the server.
+    connection = sdk.Connection(
+        url = params['output_conn'],
+        username = username,
+        password = password,
+        ca_file = params['rhv_cafile'],
+        log = logging.getLogger(),
+        insecure = params['insecure'],
+    )
+
+    system_service = connection.system_service()
+
+    # Create the disk.
+    disks_service = system_service.disks_service()
+    if params['disk_format'] == "raw":
+        disk_format = types.DiskFormat.RAW
+    else:
+        disk_format = types.DiskFormat.COW
+    disk = disks_service.add(
+        disk = types.Disk(
+            name = params['disk_name'],
+            description = "Uploaded by virt-v2v",
+            format = disk_format,
+            provisioned_size = params['disk_size'],
+            sparse = params['output_sparse'],
+            storage_domains = [
+                types.StorageDomain(
+                    name = params['output_storage'],
+                )
+            ],
+        )
+    )
+
+    # Wait till the disk is up, as the transfer can't start if the
+    # disk is locked:
+    disk_service = disks_service.disk_service(disk.id)
+
+    endt = time.time() + timeout
+    while True:
+        time.sleep(5)
+        disk = disk_service.get()
+        if disk.status == types.DiskStatus.OK:
+            break
+        if time.time() > endt:
+            raise RuntimeError("timed out waiting for disk to become
unlocked")
+
+    # Get a reference to the transfer service.
+    transfers_service = system_service.image_transfers_service()
+
+    # Create a new image transfer.
+    transfer = transfers_service.add(
+        types.ImageTransfer(
+            image = types.Image(
+                id = disk.id
+            )
+        )
+    )
+
+    # Get a reference to the created transfer service.
+    transfer_service = transfers_service.image_transfer_service(transfer.id)
+
+    # After adding a new transfer for the disk, the transfer's status
+    # will be INITIALIZING.  Wait until the init phase is over. The
+    # actual transfer can start when its status is "Transferring".
+    endt = time.time() + timeout
+    while True:
+        time.sleep(5)
+        transfer = transfer_service.get()
+        if transfer.phase != types.ImageTransferPhase.INITIALIZING:
+            break
+        if time.time() > endt:
+            raise RuntimeError("timed out waiting for transfer status !=
INITIALIZING")
+
+    # Now we have permission to start the transfer.
+    if params['rhv_direct']:
+        if transfer.transfer_url is None:
+            raise RuntimeError("direct upload to host not supported,
requires ovirt-engine >= 4.2 and only works when virt-v2v is run within the
oVirt/RHV environment, eg. on an ovirt node.")
+        destination_url = urlparse(transfer.transfer_url)
+    else:
+        destination_url = urlparse(transfer.proxy_url)
+
+    context = ssl.create_default_context()
+    context.load_verify_locations(cafile = params['rhv_cafile'])
+
+    http = HTTPSConnection(
+        destination_url.hostname,
+        destination_url.port,
+        context = context
+    )
+
+    # Save everything we need to make requests in the handle.
+    return {
+        'connection': connection,
+        'disk': disk,
+        'disk_service': disk_service,
+        'failed': False,
+        'got_options': False,
+        'highestwrite': 0,
+        'http': http,
+        'path': destination_url.path,
+        'transfer': transfer,
+        'transfer_service': transfer_service,
+    }
+
+# Can we issue zero, trim or flush requests?
+def get_options(h):
+    if h['got_options']:
+        return
+    h['got_options'] = True
+
+    http = h['http']
+    transfer=h['transfer']
+
+    http.putrequest("OPTIONS", h['path'])
+    http.putheader("Authorization", transfer.signed_ticket)
+    http.endheaders()
+
+    r = http.getresponse()
+    if r.status == 200:
+        j = json.loads(r.read())
+        h['can_zero'] = "zero" in j['features']
+        h['can_trim'] = "trim" in j['features']
+        h['can_flush'] = "flush" in j['features']
+
+    # Old imageio servers returned 405 Method Not Allowed.  If
+    # we see that we just say that no operations are allowed and
+    # we will emulate them.
+    elif r.status == 405:
+        h['can_zero'] = False
+        h['can_trim'] = False
+        h['can_flush'] = False
+
+    else:
+        raise RuntimeError("could not use OPTIONS request: %d: %s" %
+                           (r.status, r.reason))
+
+def can_trim(h):
+    get_options(h)
+    return h['can_trim']
+
+def can_flush(h):
+    get_options(h)
+    return h['can_flush']
+
+def get_size(h):
+    return params['disk_size']
+
+# For documentation see:
+# https://github.com/oVirt/ovirt-imageio/blob/master/docs/random-io.md
+# For examples of working code to read/write from the server, see:
+# https://github.com/oVirt/ovirt-imageio/blob/master/daemon/test/server_test.py
+
+def pread(h, count, offset):
+    http = h['http']
+    transfer=h['transfer']
+    transfer_service=h['transfer_service']
+
+    http.putrequest("GET", h['path'])
+    http.putheader("Authorization", transfer.signed_ticket)
+    http.putheader("Range", "bytes=%d-%d" % (offset,
offset+count-1))
+    http.endheaders()
+
+    r = http.getresponse()
+    # 206 = HTTP Partial Content, which is the usual response.
+    if r.status != 200 and r.status != 206:
+        h['transfer_service'].pause()
+        h['failed'] = True
+        raise RuntimeError("could not read sector (%d, %d): %d: %s" %
+                           (offset, count, r.status, r.reason))
+    return r.read()
+
+def pwrite(h, buf, offset):
+    count = len(buf)
+    h['highestwrite'] = max(h['highestwrite'], offset+count)
+    real_pwrite(h, buf, offset, count)
+
+def real_pwrite(h, buf, offset, count):
+    http = h['http']
+    transfer=h['transfer']
+    transfer_service=h['transfer_service']
+
+    http.putrequest("PUT", h['path'])
+    http.putheader("Authorization", transfer.signed_ticket)
+    # The oVirt server only uses the first part of the range, and the
+    # content-length.
+    http.putheader("Content-Range", "bytes %d-%d/*" %
(offset, offset+count-1))
+    http.putheader("Content-Length", str(count))
+    http.endheaders()
+    http.send(buf)
+
+    r = http.getresponse()
+    if r.status != 200:
+        transfer_service.pause()
+        h['failed'] = True
+        raise RuntimeError("could not write sector (%d, %d): %d: %s"
%
+                           (offset, count, r.status, r.reason))
+
+def zero(h, count, offset, may_trim):
+    http = h['http']
+    transfer=h['transfer']
+    transfer_service=h['transfer_service']
+
+    # Unlike the trim and flush calls, there is no 'can_zero' method
+    # so nbdkit could call this even if the server doesn't support
+    # zeroing.  If this is the case we must emulate.
+    if not h['can_zero']:
+        emulate_zero_with_pwrite(h, count, offset)
+        return
+
+    # Construct the JSON request for zeroing.
+    buf = json.dumps({'op': "zero",
+                      'offset': offset,
+                      'size': count,
+                      'flush': False})
+
+    http.putrequest("PATCH", h['path'])
+    http.putheader("Content-Type", "application/json")
+    http.putheader("Authorization", transfer.signed_ticket)
+    http.putheader("Content-Length", len(buf))
+    http.endheaders()
+    http.send(buf)
+
+    r = http.getresponse()
+    if r.status != 200:
+        transfer_service.pause()
+        h['failed'] = True
+        raise RuntimeError("could not zero sector (%d, %d): %d: %s" %
+                           (offset, count, r.status, r.reason))
+
+# qemu-img convert starts by trying to zero/trim the whole device.
+# Since we've just created a new disk it's safe to ignore these
+# requests as long as they are smaller than the highest write seen.
+# After that we must emulate them with writes.
+def emulate_zero_with_pwrite(h, count, offset):
+    if offset+count < h['highestwrite']:
+        # count could be very large, so split into chunks.
+        while count > 0:
+            n = min(count, 65536)
+            buf = bytearray(n)
+            real_pwrite(h, buf, offset, n)
+            offset += n
+            count -= n
+
+def trim(h, count, offset):
+    http = h['http']
+    transfer=h['transfer']
+    transfer_service=h['transfer_service']
+
+    # Construct the JSON request for trimming.
+    buf = json.dumps({'op': "trim",
+                      'offset': offset,
+                      'size': count,
+                      'flush': False})
+
+    http.putrequest("PATCH", h['path'])
+    http.putheader("Content-Type", "application/json")
+    http.putheader("Authorization", transfer.signed_ticket)
+    http.putheader("Content-Length", len(buf))
+    http.endheaders()
+    http.send(buf)
+
+    r = http.getresponse()
+    if r.status != 200:
+        transfer_service.pause()
+        h['failed'] = True
+        raise RuntimeError("could not trim sector (%d, %d): %d: %s" %
+                           (offset, count, r.status, r.reason))
+
+def flush(h):
+    http = h['http']
+    transfer=h['transfer']
+    transfer_service=h['transfer_service']
+
+    # Construct the JSON request for flushing.  Note the offset
+    # and count must be present but are ignored by imageio.
+    buf = json.dumps({'op': "flush",
+                      'offset': 0,
+                      'size': 0,
+                      'flush': True})
+
+    http.putrequest("PATCH", h['path'])
+    http.putheader("Content-Type", "application/json")
+    http.putheader("Authorization", transfer.signed_ticket)
+    http.putheader("Content-Length", len(buf))
+    http.endheaders()
+    http.send(buf)
+
+    r = http.getresponse()
+    if r.status != 200:
+        transfer_service.pause()
+        h['failed'] = True
+        raise RuntimeError("could not flush: %d: %s" % (r.status,
r.reason))
+
+def close(h):
+    http = h['http']
+    connection = h['connection']
+
+    http.close()
+
+    # If we didn't fail, then finalize the transfer.
+    if not h['failed']:
+        disk = h['disk']
+        transfer_service=h['transfer_service']
+
+        transfer_service.finalize()
+
+        # Wait until the transfer disk job is completed since
+        # only then we can be sure the disk is unlocked.  As this
+        # code is not very clear, what's happening is that we are
+        # waiting for the transfer object to cease to exist, which
+        # falls through to the exception case and then we can
+        # continue.
+        endt = time.time() + timeout
+        try:
+            while True:
+                time.sleep(1)
+                tmp = transfer_service.get()
+                if time.time() > endt:
+                    raise RuntimeError("timed out waiting for transfer to
finalize")
+        except sdk.NotFoundError:
+            pass
+
+        # Write the disk ID file.  Only do this on successful completion.
+        with builtins.open(params['diskid_file'], 'w') as fp:
+            fp.write(disk.id)
+
+    # Otherwise if we did fail then we should delete the disk.
+    else:
+        disk_service = h['disk_service']
+        disk_service.remove()
+
+    connection.close()
diff --git a/v2v/rhv-upload-precheck.py b/v2v/rhv-upload-precheck.py
new file mode 100644
index 000000000..e3fa65148
--- /dev/null
+++ b/v2v/rhv-upload-precheck.py
@@ -0,0 +1,72 @@
+# -*- python -*-
+# oVirt or RHV pre-upload checks used by ‘virt-v2v -o rhv-upload’
+# Copyright (C) 2018 Red Hat Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License along
+# with this program; if not, write to the Free Software Foundation, Inc.,
+# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+import json
+import logging
+import ovirtsdk4 as sdk
+import ovirtsdk4.types as types
+import sys
+import time
+
+from http.client import HTTPSConnection
+from urllib.parse import urlparse
+
+# Parameters are passed in via a JSON doc from the OCaml code.
+# Because this Python code ships embedded inside virt-v2v there
+# is no formal API here.
+params = None
+
+if len(sys.argv) != 2:
+    raise RuntimeError("incorrect number of parameters")
+
+# Parameters are passed in via a JSON document.
+with open(sys.argv[1], 'r') as fp:
+    params = json.load(fp)
+
+# What is passed in is a password file, read the actual password.
+with open(params['output_password'], 'r') as fp:
+    output_password = fp.read()
+output_password = output_password.rstrip()
+
+# Parse out the username from the output_conn URL.
+parsed = urlparse(params['output_conn'])
+username = parsed.username or "admin@internal"
+
+# Connect to the server.
+connection = sdk.Connection(
+    url = params['output_conn'],
+    username = username,
+    password = output_password,
+    ca_file = params['rhv_cafile'],
+    log = logging.getLogger(),
+    insecure = params['insecure'],
+)
+
+system_service = connection.system_service()
+
+# Find if a virtual machine already exists with that name.
+vms_service = system_service.vms_service()
+vms = vms_service.list(
+    search = ("name=%s" % params['output_name']),
+)
+if len(vms) > 0:
+    vm = vms[0]
+    raise RuntimeError("VM already exists with name ‘%s’, id ‘%s’" %
+                       (params['output_name'], vm.id))
+
+# Otherwise everything is OK, exit with no error.
diff --git a/v2v/test-v2v-python-syntax.sh b/v2v/test-v2v-python-syntax.sh
new file mode 100755
index 000000000..11ecd1f84
--- /dev/null
+++ b/v2v/test-v2v-python-syntax.sh
@@ -0,0 +1,27 @@
+#!/bin/bash -
+# libguestfs
+# Copyright (C) 2018 Red Hat Inc.
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation; either version 2 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program; if not, write to the Free Software
+# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+set -e
+
+$TEST_FUNCTIONS
+skip_if_skipped
+
+# Checks the files are syntactically correct, but not very much else.
+python3 -m py_compile rhv-upload-createvm.py
+python3 -m py_compile rhv-upload-plugin.py
+python3 -m py_compile rhv-upload-precheck.py
diff --git a/v2v/virt-v2v.pod b/v2v/virt-v2v.pod
index 412fa0579..bc6631c3c 100644
--- a/v2v/virt-v2v.pod
+++ b/v2v/virt-v2v.pod
@@ -6,15 +6,18 @@ virt-v2v - Convert a guest to use KVM
 
  virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest
 
- virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
-   -o rhv -os rhv.nfs:/export_domain --network ovirtmgmt
-
  virt-v2v -i libvirtxml guest-domain.xml -o local -os /var/tmp
 
  virt-v2v -i disk disk.img -o local -os /var/tmp
 
  virt-v2v -i disk disk.img -o glance
 
+ virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
+   -o rhv-upload -oc https://ovirt-engine.example.com/ovirt-engine/api \
+   -os ovirt-data -op /tmp/ovirt-admin-password \
+   --rhv-cafile /tmp/ca.pem --rhv-direct \
+   --network ovirtmgmt
+
  virt-v2v -ic qemu:///system qemu_guest --in-place
 
 =head1 DESCRIPTION
@@ -42,9 +45,9 @@ libguestfs E<ge> 1.28.
  Xen ───▶│ -i libvirt ──▶ │            │     │  (default) │
  ... ───▶│  (default) │   │            │ ──┐ └────────────┘
          └────────────┘   │            │ ─┐└──────▶ -o glance
- -i libvirtxml ─────────▶ │            │ ┐└─────────▶ -o rhv
- -i vmx ────────────────▶ │            │ └──────────▶ -o vdsm
-                          └────────────┘
+ -i libvirtxml ─────────▶ │            │ ┐├─────────▶ -o rhv
+ -i vmx ────────────────▶ │            │ │└─────────▶ -o vdsm
+                          └────────────┘ └──────────▶ -o rhv-upload
 
 Virt-v2v has a number of possible input and output modes, selected
 using the I<-i> and I<-o> options.  Only one input and output mode
can
@@ -103,20 +106,18 @@ For more information see L</INPUT FROM VMWARE VCENTER
SERVER> below.
 =head2 Convert from VMware to RHV/oVirt
 
 This is the same as the previous example, except you want to send the
-guest to a RHV-M Export Storage Domain which is located remotely
-(over NFS) at C<rhv.nfs:/export_domain>.  If you are unclear about
-the location of the Export Storage Domain you should check the
-settings on your RHV-M management console.  Guest network
+guest to a RHV Data Domain using the RHV REST API.  Guest network
 interface(s) are connected to the target network called C<ovirtmgmt>.
 
  virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
-   -o rhv -os rhv.nfs:/export_domain --network ovirtmgmt
+   -o rhv-upload -oc https://ovirt-engine.example.com/ovirt-engine/api \
+   -os ovirt-data -op /tmp/ovirt-admin-password \
+   --rhv-cafile /tmp/ca.pem --rhv-direct \
+   --network ovirtmgmt
 
 In this case the host running virt-v2v acts as a B<conversion server>.
 
-Note that after conversion, the guest will appear in the RHV-M Export
-Storage Domain, from where you will need to import it using the RHV-M
-user interface.  (See L</OUTPUT TO RHV>).
+For more information see L</OUTPUT TO RHV> below.
 
 =head2 Convert from ESXi hypervisor over SSH to local libvirt
 
@@ -509,6 +510,10 @@ written.
 
 This is the same as I<-o rhv>.
 
+=item B<-o> B<ovirt-upload>
+
+This is the same as I<-o rhv-upload>.
+
 =item B<-o> B<qemu>
 
 Set the output method to I<qemu>.
@@ -533,6 +538,16 @@ I<-os> parameter must also be used to specify the
location of the
 Export Storage Domain.  Note this does not actually import the guest
 into RHV.  You have to do that manually later using the UI.
 
+See L</OUTPUT TO RHV (OLD METHOD)> below.
+
+=item B<-o> B<rhv-upload>
+
+Set the output method to I<rhv-upload>.
+
+The converted guest is written directly to a RHV Data Domain.
+This is a faster method than I<-o rhv>, but requires oVirt
+or RHV E<ge> 4.2.
+
 See L</OUTPUT TO RHV> below.
 
 =item B<-o> B<vdsm>
@@ -632,6 +647,32 @@ virt-v2v finishes.
 
 This disables progress bars and other unnecessary output.
 
+=item B<--rhv-cafile> F<ca.pem>
+
+For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, the F<ca.pem>
file
+(Certificate Authority), copied from F</etc/pki/ovirt-engine/ca.pem>
+on the oVirt engine.
+
+=item B<--rhv-cluster> C<CLUSTERNAME>
+
+For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, set the RHV Cluster
+Name.  If not given it uses C<Default>.
+
+=item B<--rhv-direct>
+
+For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, if this option is
given
+then virt-v2v will attempt to directly upload the disk to the oVirt
+node, otherwise it will proxy the upload through the oVirt engine.
+Direct upload requires that you have network access to the oVirt
+nodes.  Non-direct upload is slightly slower but should work in all
+situations.
+
+=item B<--rhv-verifypeer>
+
+For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, verify the oVirt/RHV
+server’s identity by checking the server‘s certificate against the
+Certificate Authority.
+
 =item B<--root ask>
 
 =item B<--root single>
@@ -1875,6 +1916,62 @@ Define the final guest in libvirt:
 
 =head1 OUTPUT TO RHV
 
+This new method to upload guests to oVirt or RHV directly via the REST
+API requires oVirt/RHV E<ge> 4.2.
+
+You need to specify I<-o rhv-upload> as well as the following extra
+parameters:
+
+=over 4
+
+=item I<-oc> C<https://ovirt-engine.example.com/ovirt-engine/api>
+
+The URL of the REST API which is usually the server name with
+C</ovirt-engine/api> appended, but might be different if you installed
+oVirt Engine on a different path.
+
+You can optionally add a username and port number to the URL.  If the
+username is not specified then virt-v2v defaults to using
+C<admin@internal> which is the typical superuser account for oVirt
+instances.
+
+=item I<-op> F<password-file>
+
+A file containing a password to be used when connecting to the oVirt
+engine.  Note the file should contain the whole password, B<without
+any trailing newline>, and for security the file should have mode
+C<0600> so that others cannot read it.
+
+=item I<-os> C<ovirt-data>
+
+The storage domain.
+
+=item I<--rhv-cafile> F<ca.pem>
+
+The F<ca.pem> file (Certificate Authority), copied from
+F</etc/pki/ovirt-engine/ca.pem> on the oVirt engine.
+
+=item B<--rhv-cluster> C<CLUSTERNAME>
+
+Set the RHV Cluster Name.  If not given it uses C<Default>.
+
+=item I<--rhv-direct>
+
+If this option is given then virt-v2v will attempt to directly upload
+the disk to the oVirt node, otherwise it will proxy the upload through
+the oVirt engine.  Direct upload requires that you have network access
+to the oVirt nodes.  Non-direct upload is slightly slower but should
+work in all situations.
+
+=item I<--rhv-verifypeer>
+
+Verify the oVirt/RHV server’s identity by checking the server‘s
+certificate against the Certificate Authority.
+
+=back
+
+=head1 OUTPUT TO RHV (OLD METHOD)
+
 This section only applies to the I<-o rhv> output mode.  If you use
 virt-v2v from the RHV-M user interface, then behind the scenes the
 import is managed by VDSM using the I<-o vdsm> output mode (which end
-- 
2.13.2

On Wed, Mar 21, 2018 at 3:34 PM Richard W.M. Jones <rjones@redhat.com>
wrote:
> PROBLEMS:
>  - -of qcow2 does not work, with multiple problems
>     * needs to set NBD size to something larger than virtual size
>
> This adds a new output mode to virt-v2v.  virt-v2v -o rhv-upload
> streams images directly to an oVirt or RHV >= 4 Data Domain using the
> oVirt SDK v4.  It is more efficient than -o rhv because it does not
> need to go via the Export Storage Domain, and is possible for humans
> to use unlike -o vdsm.
>
> The implementation uses the Python SDK (‘ovirtsdk4’ module).  An
> nbdkit Python 3 plugin translates NBD calls from qemu into HTTPS
> requests to oVirt via the SDK.
> ---
>  .gitignore                                |   3 +
>  v2v/Makefile.am                           |  28 ++-
>  v2v/cmdline.ml                            |  46 ++++
>  v2v/embed.sh                              |  45 ++++
>  v2v/output_rhv_upload.ml                  | 339 +++++++++++++++++++++++++
>  v2v/output_rhv_upload.mli                 |  27 ++
>  v2v/output_rhv_upload_createvm_source.mli |  19 ++
>  v2v/output_rhv_upload_plugin_source.mli   |  19 ++
>  v2v/output_rhv_upload_precheck_source.mli |  19 ++
>  v2v/rhv-upload-createvm.py                |  85 +++++++
>  v2v/rhv-upload-plugin.py                  | 394
> ++++++++++++++++++++++++++++++
>  v2v/rhv-upload-precheck.py                |  72 ++++++
>  v2v/test-v2v-python-syntax.sh             |  27 ++
>  v2v/virt-v2v.pod                          | 125 ++++++++--
>  14 files changed, 1233 insertions(+), 15 deletions(-)
>
> diff --git a/.gitignore b/.gitignore
> index d72447d1d..211376eef 100644
> --- a/.gitignore
> +++ b/.gitignore
> @@ -654,6 +654,9 @@ Makefile.in
>  /utils/qemu-speed-test/qemu-speed-test
>  /v2v/.depend
>  /v2v/oUnit-*
> +/v2v/output_rhv_upload_createvm_source.ml
> +/v2v/output_rhv_upload_plugin_source.ml
> +/v2v/output_rhv_upload_precheck_source.ml
>  /v2v/real-*.d/
>  /v2v/real-*.img
>  /v2v/real-*.xml
> diff --git a/v2v/Makefile.am b/v2v/Makefile.am
> index c2eb31097..987ad8319 100644
> --- a/v2v/Makefile.am
> +++ b/v2v/Makefile.am
> @@ -22,12 +22,19 @@ generator_built = \
>         uefi.mli
>
>  BUILT_SOURCES = \
> -       $(generator_built)
> +       $(generator_built) \
> +       output_rhv_upload_createvm_source.ml \
> +       output_rhv_upload_plugin_source.ml \
> +       output_rhv_upload_precheck_source.ml
>
>  EXTRA_DIST = \
>         $(SOURCES_MLI) $(SOURCES_ML) $(SOURCES_C) \
>         copy_to_local.ml \
>         copy_to_local.mli \
> +       embed-code.sh \
> +       rhv-upload-createvm.py \
> +       rhv-upload-plugin.py \
> +       rhv-upload-precheck.py \
>         v2v_slow_unit_tests.ml \
>         v2v-slow-unit-tests.sh \
>         v2v_unit_tests.ml \
> @@ -64,6 +71,10 @@ SOURCES_MLI = \
>         output_null.mli \
>         output_qemu.mli \
>         output_rhv.mli \
> +       output_rhv_upload.mli \
> +       output_rhv_upload_createvm_source.mli \
> +       output_rhv_upload_plugin_source.mli \
> +       output_rhv_upload_precheck_source.mli \
>         output_vdsm.mli \
>         parse_ovf_from_ova.mli \
>         parse_libvirt_xml.mli \
> @@ -116,6 +127,10 @@ SOURCES_ML = \
>         output_local.ml \
>         output_qemu.ml \
>         output_rhv.ml \
> +       output_rhv_upload_createvm_source.ml \
> +       output_rhv_upload_plugin_source.ml \
> +       output_rhv_upload_precheck_source.ml \
> +       output_rhv_upload.ml \
>         output_vdsm.ml \
>         inspect_source.ml \
>         target_bus_assignment.ml \
> @@ -126,6 +141,15 @@ SOURCES_C = \
>         libvirt_utils-c.c \
>         qemuopts-c.c
>
> +# These files are generated and contain rhv-upload-*.py embedded as an
> +# OCaml string.
> +output_rhv_upload_createvm_source.ml: rhv-upload-createvm.py
> +       ./embed.sh code $^ $@
> +output_rhv_upload_plugin_source.ml: rhv-upload-plugin.py
> +       ./embed.sh code $^ $@
> +output_rhv_upload_precheck_source.ml: rhv-upload-precheck.py
> +       ./embed.sh code $^ $@
> +
>  if HAVE_OCAML
>
>  bin_PROGRAMS = virt-v2v virt-v2v-copy-to-local
> @@ -295,6 +319,7 @@ TESTS_ENVIRONMENT = $(top_builddir)/run --test
>
>  TESTS = \
>         test-v2v-docs.sh \
> +       test-v2v-python-syntax.sh \
>         test-v2v-i-ova-bad-sha1.sh \
>         test-v2v-i-ova-bad-sha256.sh \
>         test-v2v-i-ova-formats.sh \
> @@ -490,6 +515,7 @@ EXTRA_DIST += \
>         test-v2v-print-source.expected \
>         test-v2v-print-source.sh \
>         test-v2v-print-source.xml \
> +       test-v2v-python-syntax.sh \
>         test-v2v-conversion-of.sh \
>         test-v2v-sound.sh \
>         test-v2v-sound.xml \
> diff --git a/v2v/cmdline.ml b/v2v/cmdline.ml
> index 14c5c0dca..1b3c10be8 100644
> --- a/v2v/cmdline.ml
> +++ b/v2v/cmdline.ml
> @@ -67,6 +67,10 @@ let parse_cmdline () >    let output_password = ref
None in
>    let output_storage = ref None in
>    let password_file = ref None in
> +  let rhv_cafile = ref None in
> +  let rhv_cluster = ref None in
> +  let rhv_direct = ref false in
> +  let rhv_verifypeer = ref false in
>    let vddk_config = ref None in
>    let vddk_cookie = ref None in
>    let vddk_libdir = ref None in
> @@ -145,6 +149,8 @@ let parse_cmdline () >      | "disk" |
"local" -> output_mode := `Local
>      | "null" -> output_mode := `Null
>      | "ovirt" | "rhv" | "rhev" ->
output_mode := `RHV
> +    | "ovirt-upload" | "ovirt_upload" |
"rhv-upload" | "rhv_upload" ->
> +       output_mode := `RHV_Upload
>      | "qemu" -> output_mode := `QEmu
>      | "vdsm" -> output_mode := `VDSM
>      | s ->
> @@ -233,6 +239,12 @@ let parse_cmdline () >      [
L"print-target" ], Getopt.Set print_target,
>                                      s_"Print target and stop";
>      [ L"qemu-boot" ], Getopt.Set qemu_boot, s_"Boot in qemu
(-o qemu
> only)";
> +    [ L"rhv-cafile" ], Getopt.String ("ca.pem",
set_string_option_once
> "--rhv-cafile" rhv_cafile),
> +                                    s_"For -o rhv-upload, set
‘ca.pem’
> file";
> +    [ L"rhv-cluster" ], Getopt.String ("clustername",
> set_string_option_once "--rhv-cluster" rhv_cluster),
> +                                    s_"For -o rhv-upload, set cluster
> name";
> +    [ L"rhv-direct" ], Getopt.Set rhv_direct, s_"Use direct
transfer
> mode";
> +    [ L"rhv-verifypeer" ], Getopt.Set rhv_verifypeer,
s_"Verify server
> identity";
>      [ L"root" ],     Getopt.String ("ask|... ",
set_root_choice),
>                                      s_"How to choose root
filesystem";
>      [ L"vddk-config" ], Getopt.String ("filename",
set_string_option_once
> "--vddk-config" vddk_config),
> @@ -327,6 +339,10 @@ read the man page virt-v2v(1).
>    let print_source = !print_source in
>    let print_target = !print_target in
>    let qemu_boot = !qemu_boot in
> +  let rhv_cafile = !rhv_cafile in
> +  let rhv_cluster = !rhv_cluster in
> +  let rhv_direct = !rhv_direct in
> +  let rhv_verifypeer = !rhv_verifypeer in
>    let root_choice = !root_choice in
>    let vddk_options >        { vddk_config = !vddk_config;
> @@ -557,6 +573,36 @@ read the man page virt-v2v(1).
>        Output_rhv.output_rhv os output_alloc,
>        output_format, output_alloc
>
> +    | `RHV_Upload ->
> +      let output_conn > +        match output_conn with
> +        | None ->
> +           error (f_"-o rhv-upload: use ‘-oc’ to point to the oVirt
or
> RHV server REST API URL, which is usually
> https://servername/ovirt-engine/api")
> +        | Some oc -> oc in
> +      (* In theory we could make the password optional in future. *)
> +      let output_password > +        match output_password with
> +        | None ->
> +           error (f_"-o rhv-upload: output password file was not
> specified, use ‘-op’ to point to a file which contains the password used to
> connect to the oVirt or RHV server")
> +        | Some op -> op in
> +      let os > +        match output_storage with
> +        | None ->
> +           error (f_"-o rhv-upload: output storage was not specified,
use
> ‘-os’");
> +        | Some os -> os in
> +      if qemu_boot then
> +        error_option_cannot_be_used_in_output_mode "rhv-upload"
> "--qemu-boot";
> +      let rhv_cafile > +        match rhv_cafile with
> +        | None ->
> +           error (f_"-o rhv-upload: must use ‘--rhv-cafile’ to supply
the
> path to the oVirt or RHV server’s ‘ca.pem’ file")
> +        | Some rhv_cafile -> rhv_cafile in
> +      Output_rhv_upload.output_rhv_upload output_alloc output_conn
> +                                          output_password os
> +                                          rhv_cafile rhv_cluster
> +                                          rhv_direct rhv_verifypeer,
> +      output_format, output_alloc
> +
>      | `VDSM ->
>        if output_password <> None then
>          error_option_cannot_be_used_in_output_mode "vdsm"
"-op";
> diff --git a/v2v/embed.sh b/v2v/embed.sh
> new file mode 100755
> index 000000000..0a65cd428
> --- /dev/null
> +++ b/v2v/embed.sh
> @@ -0,0 +1,45 @@
> +#!/bin/bash -
> +# Embed code or other content into an OCaml file.
> +# Copyright (C) 2018 Red Hat Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free&entry=gmail&source=g>
> Software
> +# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> 02110-1301 USA.
> +
> +# Embed code or other content into an OCaml file.
> +#
> +# It is embedded into a string.  As OCaml string literals have virtually
> +# no restrictions on length or content we only have to escape double
> +# quotes for backslash characters.
> +
> +if [ $# -ne 3 ]; then
> +    echo "embed.sh identifier input output"
> +    exit 1
> +fi
> +
> +ident="$1"
> +input="$2"
> +output="$3"
> +
> +rm -f "$output" "$output"-t
> +
> +exec >"$output"-t
> +
> +echo "(* Generated by embed.sh from $input *)"
> +echo
> +echo let "$ident" = '"'
> +sed -e 's/\(["\]\)/\\\1/g' < "$input"
> +echo '"'
> +
> +chmod -w "$output"-t
> +mv "$output"-t "$output"
> diff --git a/v2v/output_rhv_upload.ml b/v2v/output_rhv_upload.ml
> new file mode 100644
> index 000000000..77f0afa40
> --- /dev/null
> +++ b/v2v/output_rhv_upload.ml
> @@ -0,0 +1,339 @@
> +(* virt-v2v
> + * Copyright (C) 2009-2018 Red Hat Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free+&entry=gmail&source=g>Software
> Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + *)
> +
> +open Printf
> +open Unix
> +
> +open Std_utils
> +open Tools_utils
> +open Unix_utils
> +open Common_gettext.Gettext
> +
> +open Types
> +open Utils
> +
> +let python3 = "python3" (* Defined by PEP 394 *)
> +let pidfile_timeout = 30
> +let finalization_timeout = 5*60
> +
> +class output_rhv_upload output_alloc output_conn
> +                        output_password output_storage
> +                        rhv_cafile rhv_cluster
> +                        rhv_direct rhv_verifypeer > +  (* Create a
temporary directory which will be deleted on exit. *)
> +  let tmpdir > +    let base_dir = (open_guestfs ())#get_cachedir () in
> +    let t = Mkdtemp.temp_dir ~base_dir "rhvupload." in
> +    rmdir_on_exit t;
> +    t in
> +
> +  let diskid_file_of_id id = tmpdir // sprintf "diskid.%d" id in
> +
> +  (* Write the Python precheck, plugin and create VM to a temporary file.
> *)
> +  let precheck > +    let precheck = tmpdir //
"rhv-upload-precheck.py" in
> +    with_open_out
> +      precheck
> +      (fun chan -> output_string chan
> Output_rhv_upload_precheck_source.code);
> +    precheck in
> +  let plugin > +    let plugin = tmpdir //
"rhv-upload-plugin.py" in
> +    with_open_out
> +      plugin
> +      (fun chan -> output_string chan
> Output_rhv_upload_plugin_source.code);
> +    plugin in
> +  let createvm > +    let createvm = tmpdir //
"rhv-upload-createvm.py" in
> +    with_open_out
> +      createvm
> +      (fun chan -> output_string chan
> Output_rhv_upload_createvm_source.code);
> +    createvm in
> +
> +  (* Is SELinux enabled and enforcing on the host? *)
> +  let have_selinux > +    0 = Sys.command "getenforce
2>/dev/null | grep -isq Enforcing" in
> +
> +  (* Check that nbdkit is available and new enough. *)
> +  let error_unless_nbdkit_working () > +    if 0 <> Sys.command
"nbdkit --version >/dev/null" then
> +      error (f_"nbdkit is not installed or not working.  It is
required
> to use ‘-o rhv-upload’.  See \"OUTPUT TO RHV\" in the virt-v2v(1)
manual.");
> +
> +    (* Check it's a new enough version.  The latest features we
> +     * require are ‘--exit-with-parent’ and ‘--selinux-label’, both
> +     * added in 1.1.14.  (We use 1.1.16 as the minimum here because
> +     * it also adds the selinux=yes|no flag in --dump-config).
> +     *)
> +    let lines = external_command "nbdkit --help" in
> +    let lines = String.concat " " lines in
> +    if String.find lines "exit-with-parent" == -1 ||
> +       String.find lines "selinux-label" == -1 then
> +      error (f_"nbdkit is not new enough, you need to upgrade to
nbdkit ≥
> 1.1.16")
> +  in
> +
> +  (* Check that the python3 plugin is installed and working
> +   * and can load the plugin script.
> +   *)
> +  let error_unless_nbdkit_python3_working () > +    let cmd = sprintf
"nbdkit %s %s --dump-plugin >/dev/null"
> +                      python3 (quote plugin) in
> +    if Sys.command cmd <> 0 then
> +      error (f_"nbdkit Python 3 plugin is not installed or not
working.
> It is required if you want to use ‘-o rhv-upload’.
> +
> +See also \"OUTPUT TO RHV\" in the virt-v2v(1) manual.")
> +  in
> +
> +  (* Check that nbdkit was compiled with SELinux support (for the
> +   * --selinux-label option).
> +   *)
> +  let error_unless_nbdkit_compiled_with_selinux () > +    let lines =
external_command "nbdkit --dump-config" in
> +    (* In nbdkit <= 1.1.15 the selinux attribute was not present
> +     * at all in --dump-config output so there was no way to tell.
> +     * Ignore this case because there will be an error later when
> +     * we try to use the --selinux-label parameter.
> +     *)
> +    if List.mem "selinux=no" (List.map String.trim lines) then
> +      error (f_"nbdkit was compiled without SELinux support.  You
will
> have to recompile nbdkit with libselinux-devel installed, or else set
> SELinux to Permissive mode while doing the conversion.")
> +  in
> +
> +  (* JSON parameters which are invariant between disks. *)
> +  let json_params = [
> +    "output_conn", JSON.String output_conn;
> +    "output_password", JSON.String output_password;
> +    "output_storage", JSON.String output_storage;
> +    "output_sparse", JSON.Bool (match output_alloc with
> +                                | Sparse -> true
> +                                | Preallocated -> false);
> +    "rhv_cafile", JSON.String rhv_cafile;
> +    "rhv_cluster", JSON.String (Option.default
"Default" rhv_cluster);
> +    "rhv_direct", JSON.Bool rhv_direct;
> +
> +    (* The 'Insecure' flag seems to be a number with various
possible
> +     * meanings, however we just set it to True/False.
> +     *
> +     *
>
https://github.com/oVirt/ovirt-engine-sdk/blob/19aa7070b80e60a4cfd910448287aecf9083acbe/sdk/lib/ovirtsdk4/__init__.py#L395
> +     *)
> +    "insecure", JSON.Bool (not rhv_verifypeer);
> +  ] in
> +
> +  (* nbdkit command line args which are invariant between disks. *)
> +  let nbdkit_args > +    let args = [
> +      "nbdkit";
> +
> +      "--foreground";           (* run in foreground *)
> +      "--exit-with-parent";     (* exit when virt-v2v exits *)
> +      "--newstyle";             (* use newstyle NBD protocol *)
> +      "--exportname"; "/";
> +
> +      "python3";                (* use the nbdkit Python 3
plugin *)
> +      plugin;                   (* Python plugin script *)
> +    ] in
> +    let args = if verbose () then args @ ["--verbose"] else args
in
> +    let args > +      (* label the socket so qemu can open it *)
> +      if have_selinux then
> +        args @ ["--selinux-label";
"system_u:object_r:svirt_t:s0"]
> +      else args in
> +    args in
> +
> +object
> +  inherit output
> +
> +  method precheck () > +    error_unless_nbdkit_working ();
> +    error_unless_nbdkit_python3_working ();
> +    if have_selinux then
> +      error_unless_nbdkit_compiled_with_selinux ()
> +
> +  method as_options > +    "-o rhv-upload" ^
> +    (match output_alloc with
> +     | Sparse -> "" (* default, don't need to print it *)
> +     | Preallocated -> " -oa preallocated") ^
> +    sprintf " -oc %s -op %s -os %s"
> +            output_conn output_password output_storage
> +
> +  method supported_firmware = [ TargetBIOS ]
> +
> +  method prepare_targets source targets > +    let output_name =
source.s_name in
> +    let json_params > +      ("output_name", JSON.String
output_name) :: json_params in
> +
> +    (* Python code prechecks.  These can't run in #precheck because
> +     * we need to know the name of the virtual machine.
> +     *)
> +    let json_param_file = tmpdir // "params.json" in
> +    with_open_out
> +      json_param_file
> +      (fun chan -> output_string chan (JSON.string_of_doc
json_params));
> +    if run_command [ python3; precheck; json_param_file ] <> 0 then
> +      error (f_"failed server prechecks, see earlier errors");
> +
> +    (* Create an nbdkit instance for each disk and set the
> +     * target URI to point to the NBD socket.
> +     *)
> +    List.map (
> +      fun t ->
> +        let id = t.target_overlay.ov_source.s_disk_id in
> +        let disk_name = sprintf "%s-%03d" output_name id in
> +        let json_params > +          ("disk_name",
JSON.String disk_name) :: json_params in
> +
> +        let disk_format > +          match t.target_format with
> +          | ("raw" | "qcow2") as fmt -> fmt
> +          | _ ->
> +             error (f_"rhv-upload: -of %s: Only output format ‘raw’
or
> ‘qcow2’ is supported.  If the input is in a different format then force one
> of these output formats by adding either ‘-of raw’ or ‘-of qcow2’ on the
> command line.")
> +                   t.target_format in
> +        let json_params > +          ("disk_format",
JSON.String disk_format) :: json_params in
> +
> +        let disk_size = t.target_overlay.ov_virtual_size in
> +        let json_params > +          ("disk_size", JSON.Int64
disk_size) :: json_params in
> +
> +        (* Ask the plugin to write the disk ID to a special file. *)
> +        let diskid_file = diskid_file_of_id id in
> +        let json_params > +          ("diskid_file",
JSON.String diskid_file) :: json_params in
> +
> +        (* Write the JSON parameters to a file. *)
> +        let json_param_file = tmpdir // sprintf "params%d.json"
id in
> +        with_open_out
> +          json_param_file
> +          (fun chan -> output_string chan (JSON.string_of_doc
> json_params));
> +
> +        let sock = tmpdir // sprintf "nbdkit%d.sock" id in
> +        let pidfile = tmpdir // sprintf "nbdkit%d.pid" id in
> +
> +        (* Add common arguments to per-target arguments. *)
> +        let args > +          nbdkit_args @ [ "--pidfile";
pidfile;
> +                          "--unix"; sock;
> +                          sprintf "params=%s" json_param_file ]
in
> +
> +        (* Print the full command we are about to run when debugging. *)
> +        if verbose () then (
> +          eprintf "running nbdkit:\n";
> +          List.iter (fun arg -> eprintf " %s" (quote arg))
args;
> +          prerr_newline ()
> +        );
> +
> +        (* Start an nbdkit instance in the background.  By using
> +         * --exit-with-parent we don't have to worry about clean-up.
> +         *)
> +        let args = Array.of_list args in
> +        let pid = fork () in
> +        if pid = 0 then (
> +          (* Child process (nbdkit). *)
> +          execvp "nbdkit" args
> +        );
> +
> +        (* Wait for the pidfile to appear so we know that nbdkit
> +         * is listening for requests.
> +         *)
> +        if not (wait_for_file pidfile pidfile_timeout) then (
> +          if verbose () then
> +            error (f_"nbdkit did not start up.  See previous
debugging
> messages for problems.")
> +          else
> +            error (f_"nbdkit did not start up.  There may be errors
> printed by nbdkit above.
> +
> +If the messages above are not sufficient to diagnose the problem then add
> the ‘virt-v2v -v -x’ options and examine the debugging output
carefully.")
> +        );
> +
> +        if have_selinux then (
> +          (* Note that Unix domain sockets have both a file label and
> +           * a socket/process label.  Using --selinux-label above
> +           * only set the socket label, but we must also set the file
> +           * label.
> +           *)
> +          ignore (
> +              run_command ["chcon";
"system_u:object_r:svirt_image_t:s0";
> +                           sock]
> +          );
> +        );
> +        (* ... and the regular Unix permissions, in case qemu is
> +         * running as another user.
> +         *)
> +        chmod sock 0o777;
> +
> +        (* Tell ‘qemu-img convert’ to write to the nbd socket which is
> +         * connected to nbdkit.
> +         *)
> +        let json_params = [
> +          "file.driver", JSON.String "nbd";
> +          "file.path", JSON.String sock;
> +          "file.export", JSON.String "/";
> +        ] in
> +        let target_file > +          TargetURI ("json:" ^
JSON.string_of_doc json_params) in
> +        { t with target_file }
> +    ) targets
> +
> +  method create_metadata source targets _ guestcaps inspect
> target_firmware > +    (* Get the UUIDs of each disk image.  These files
are written
> +     * out by the nbdkit plugins on successful finalization of the
> +     * transfer.
> +     *)
> +    let nr_disks = List.length targets in
> +    let image_uuids > +      List.map (
> +        fun t ->
> +          let id = t.target_overlay.ov_source.s_disk_id in
> +          let diskid_file = diskid_file_of_id id in
> +          if not (wait_for_file diskid_file finalization_timeout) then
> +            error (f_"transfer of disk %d/%d failed, see earlier
error
> messages")
> +                  (id+1) nr_disks;
> +          let diskid = read_whole_file diskid_file in
> +          diskid
> +      ) targets in
> +
> +    (* We don't have the storage domain UUID, but instead we write
> +     * in a magic value which the Python code (which can get it)
> +     * will substitute.
> +     *)
> +    let sd_uuid = "@SD_UUID@" in
> +
> +    (* The volume and VM UUIDs are made up. *)
> +    let vol_uuids = List.map (fun _ -> uuidgen ()) targets
> +    and vm_uuid = uuidgen () in
> +
> +    (* Create the metadata. *)
> +    let ovf > +      Create_ovf.create_ovf source targets guestcaps
inspect
> +                            output_alloc
> +                            sd_uuid image_uuids vol_uuids vm_uuid
> +                            OVirt in
> +    let ovf = DOM.doc_to_string ovf in
> +
> +    let json_param_file = tmpdir // "params.json" in
> +    with_open_out
> +      json_param_file
> +      (fun chan -> output_string chan (JSON.string_of_doc
json_params));
> +
> +    let ovf_file = tmpdir // "vm.ovf" in
> +    with_open_out ovf_file (fun chan -> output_string chan ovf);
> +    if run_command [ python3; createvm; json_param_file; ovf_file ]
<> 0
> then
> +      error (f_"failed to create virtual machine, see earlier
errors")
> +
> +end
> +
> +let output_rhv_upload = new output_rhv_upload
> +let () = Modules_list.register_output_module "rhv-upload"
> diff --git a/v2v/output_rhv_upload.mli b/v2v/output_rhv_upload.mli
> new file mode 100644
> index 000000000..c35d83f7e
> --- /dev/null
> +++ b/v2v/output_rhv_upload.mli
> @@ -0,0 +1,27 @@
> +(* virt-v2v
> + * Copyright (C) 2009-2018 Red Hat Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free+&entry=gmail&source=g>Software
> Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + *)
> +
> +(** [-o rhv-upload] target. *)
> +
> +val output_rhv_upload : Types.output_allocation -> string -> string
->
> +                        string -> string -> string option -> bool
-> bool
> ->
> +                        Types.output
> +(** [output_rhv_upload output_alloc output_conn output_password
> output_storage
> +     rhv_cafile rhv_cluster rhv_direct rhv_verifypeer]
> +    creates and returns a new {!Types.output} object specialized for
> writing
> +    output to oVirt or RHV directly via RHV APIs. *)
> diff --git a/v2v/output_rhv_upload_createvm_source.mli
> b/v2v/output_rhv_upload_createvm_source.mli
> new file mode 100644
> index 000000000..c1bafa15b
> --- /dev/null
> +++ b/v2v/output_rhv_upload_createvm_source.mli
> @@ -0,0 +1,19 @@
> +(* virt-v2v
> + * Copyright (C) 2018 Red Hat Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free+&entry=gmail&source=g>Software
> Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + *)
> +
> +val code : string
> diff --git a/v2v/output_rhv_upload_plugin_source.mli
> b/v2v/output_rhv_upload_plugin_source.mli
> new file mode 100644
> index 000000000..c1bafa15b
> --- /dev/null
> +++ b/v2v/output_rhv_upload_plugin_source.mli
> @@ -0,0 +1,19 @@
> +(* virt-v2v
> + * Copyright (C) 2018 Red Hat Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free+&entry=gmail&source=g>Software
> Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + *)
> +
> +val code : string
> diff --git a/v2v/output_rhv_upload_precheck_source.mli
> b/v2v/output_rhv_upload_precheck_source.mli
> new file mode 100644
> index 000000000..c1bafa15b
> --- /dev/null
> +++ b/v2v/output_rhv_upload_precheck_source.mli
> @@ -0,0 +1,19 @@
> +(* virt-v2v
> + * Copyright (C) 2018 Red Hat Inc.
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License as published by
> + * the Free Software Foundation; either version 2 of the License, or
> + * (at your option) any later version.
> + *
> + * This program is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> + * GNU General Public License for more details.
> + *
> + * You should have received a copy of the GNU General Public License along
> + * with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free+&entry=gmail&source=g>Software
> Foundation, Inc.,
> + * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> + *)
> +
> +val code : string
> diff --git a/v2v/rhv-upload-createvm.py b/v2v/rhv-upload-createvm.py
> new file mode 100644
> index 000000000..7a80fa5bd
> --- /dev/null
> +++ b/v2v/rhv-upload-createvm.py
> @@ -0,0 +1,85 @@
> +# -*- python -*-
> +# oVirt or RHV upload create VM used by ‘virt-v2v -o rhv-upload’
> +# Copyright (C) 2018 Red Hat Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License along
> +# with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free&entry=gmail&source=g>
> Software Foundation, Inc.,
> +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> +
> +import json
> +import logging
> +import ovirtsdk4 as sdk
> +import ovirtsdk4.types as types
> +import sys
> +import time
> +
> +from http.client import HTTPSConnection
> +from urllib.parse import urlparse
> +
> +# Parameters are passed in via a JSON doc from the OCaml code.
> +# Because this Python code ships embedded inside virt-v2v there
> +# is no formal API here.
> +params = None
> +ovf = None                      # OVF file
> +
> +if len(sys.argv) != 3:
> +    raise RuntimeError("incorrect number of parameters")
> +
> +# Parameters are passed in via a JSON document.
> +with open(sys.argv[1], 'r') as fp:
> +    params = json.load(fp)
> +
> +# What is passed in is a password file, read the actual password.
> +with open(params['output_password'], 'r') as fp:
> +    output_password = fp.read()
> +output_password = output_password.rstrip()
> +
> +# Read the OVF document.
> +with open(sys.argv[2], 'r') as fp:
> +    ovf = fp.read()
> +
> +# Parse out the username from the output_conn URL.
> +parsed = urlparse(params['output_conn'])
> +username = parsed.username or "admin@internal"
> +
> +# Connect to the server.
> +connection = sdk.Connection(
> +    url = params['output_conn'],
> +    username = username,
> +    password = output_password,
> +    ca_file = params['rhv_cafile'],
> +    log = logging.getLogger(),
> +    insecure = params['insecure'],
> +)
> +
> +system_service = connection.system_service()
> +
> +# Get the storage domain UUID and substitute it into the OVF doc.
> +sds_service = system_service.storage_domains_service()
> +sd = sds_service.list(search=("name=%s" %
params['output_storage']))[0]
> +sd_uuid = sd.id
> +
> +ovf.replace("@SD_UUID@", sd_uuid)
> +
> +vms_service = system_service.vms_service()
> +vm = vms_service.add(
> +    types.Vm(
> +        cluster=types.Cluster(name = params['rhv_cluster']),
> +        initialization=types.Initialization(
> +            configuration = types.Configuration(
> +                type = types.ConfigurationType.OVA,
> +                data = ovf,
> +            )
> +        )
> +    )
> +)
> diff --git a/v2v/rhv-upload-plugin.py b/v2v/rhv-upload-plugin.py
> new file mode 100644
> index 000000000..0ece05b94
> --- /dev/null
> +++ b/v2v/rhv-upload-plugin.py
> @@ -0,0 +1,394 @@
> +# -*- python -*-
> +# oVirt or RHV upload nbdkit plugin used by ‘virt-v2v -o rhv-upload’
> +# Copyright (C) 2018 Red Hat Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License along
> +# with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free&entry=gmail&source=g>
> Software Foundation, Inc.,
> +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> +
> +import builtins
> +import json
> +import logging
> +import ovirtsdk4 as sdk
> +import ovirtsdk4.types as types
> +import ssl
> +import sys
> +import time
> +
> +from http.client import HTTPSConnection
> +from urllib.parse import urlparse
> +
> +# Timeout to wait for oVirt disks to change status, or the transfer
> +# object to finish initializing [seconds].
> +timeout = 5*60
> +
> +# Parameters are passed in via a JSON doc from the OCaml code.
> +# Because this Python code ships embedded inside virt-v2v there
> +# is no formal API here.
> +params = None
> +
> +def config(key, value):
> +    global params
> +
> +    if key == "params":
> +        with builtins.open(value, 'r') as fp:
> +            params = json.load(fp)
> +    else:
> +        raise RuntimeError("unknown configuration key
'%s'" % key)
> +
> +def config_complete():
> +    if params is None:
> +        raise RuntimeError("missing configuration parameters")
> +
> +def open(readonly):
> +    # Parse out the username from the output_conn URL.
> +    parsed = urlparse(params['output_conn'])
> +    username = parsed.username or "admin@internal"
> +
> +    # Read the password from file.
> +    with builtins.open(params['output_password'], 'r') as
fp:
> +        password = fp.read()
> +    password = password.rstrip()
> +
> +    # Connect to the server.
> +    connection = sdk.Connection(
> +        url = params['output_conn'],
> +        username = username,
> +        password = password,
> +        ca_file = params['rhv_cafile'],
> +        log = logging.getLogger(),
> +        insecure = params['insecure'],
> +    )
> +
> +    system_service = connection.system_service()
> +
> +    # Create the disk.
> +    disks_service = system_service.disks_service()
> +    if params['disk_format'] == "raw":
> +        disk_format = types.DiskFormat.RAW
> +    else:
> +        disk_format = types.DiskFormat.COW
> +    disk = disks_service.add(
> +        disk = types.Disk(
> +            name = params['disk_name'],
> +            description = "Uploaded by virt-v2v",
> +            format = disk_format,
> +            provisioned_size = params['disk_size'],
> +            sparse = params['output_sparse'],
> +            storage_domains = [
> +                types.StorageDomain(
> +                    name = params['output_storage'],
> +                )
> +            ],
> +        )
> +    )
> +
> +    # Wait till the disk is up, as the transfer can't start if the
> +    # disk is locked:
> +    disk_service = disks_service.disk_service(disk.id)
> +
> +    endt = time.time() + timeout
> +    while True:
> +        time.sleep(5)
> +        disk = disk_service.get()
> +        if disk.status == types.DiskStatus.OK:
> +            break
> +        if time.time() > endt:
> +            raise RuntimeError("timed out waiting for disk to become
> unlocked")
> +
> +    # Get a reference to the transfer service.
> +    transfers_service = system_service.image_transfers_service()
> +
> +    # Create a new image transfer.
> +    transfer = transfers_service.add(
> +        types.ImageTransfer(
> +            image = types.Image(
> +                id = disk.id
> +            )
> +        )
> +    )
> +
> +    # Get a reference to the created transfer service.
> +    transfer_service = transfers_service.image_transfer_service(
> transfer.id)
> +
> +    # After adding a new transfer for the disk, the transfer's status
> +    # will be INITIALIZING.  Wait until the init phase is over. The
> +    # actual transfer can start when its status is
"Transferring".
> +    endt = time.time() + timeout
> +    while True:
> +        time.sleep(5)
> +        transfer = transfer_service.get()
> +        if transfer.phase != types.ImageTransferPhase.INITIALIZING:
> +            break
> +        if time.time() > endt:
> +            raise RuntimeError("timed out waiting for transfer status
!> INITIALIZING")
> +
> +    # Now we have permission to start the transfer.
> +    if params['rhv_direct']:
> +        if transfer.transfer_url is None:
> +            raise RuntimeError("direct upload to host not supported,
> requires ovirt-engine >= 4.2 and only works when virt-v2v is run within
the
> oVirt/RHV environment, eg. on an ovirt node.")
> +        destination_url = urlparse(transfer.transfer_url)
> +    else:
> +        destination_url = urlparse(transfer.proxy_url)
> +
> +    context = ssl.create_default_context()
> +    context.load_verify_locations(cafile = params['rhv_cafile'])
> +
> +    http = HTTPSConnection(
> +        destination_url.hostname,
> +        destination_url.port,
> +        context = context
> +    )
> +
> +    # Save everything we need to make requests in the handle.
> +    return {
> +        'connection': connection,
> +        'disk': disk,
> +        'disk_service': disk_service,
> +        'failed': False,
> +        'got_options': False,
> +        'highestwrite': 0,
> +        'http': http,
> +        'path': destination_url.path,
> +        'transfer': transfer,
> +        'transfer_service': transfer_service,
> +    }
> +
> +# Can we issue zero, trim or flush requests?
> +def get_options(h):
> +    if h['got_options']:
> +        return
> +    h['got_options'] = True
> +
> +    http = h['http']
> +    transfer=h['transfer']
> +
> +    http.putrequest("OPTIONS", h['path'])
> +    http.putheader("Authorization", transfer.signed_ticket)
> +    http.endheaders()
> +
> +    r = http.getresponse()
> +    if r.status == 200:
> +        j = json.loads(r.read())
> +        h['can_zero'] = "zero" in j['features']
> +        h['can_trim'] = "trim" in j['features']
> +        h['can_flush'] = "flush" in
j['features']
> +
> +    # Old imageio servers returned 405 Method Not Allowed.  If
> +    # we see that we just say that no operations are allowed and
> +    # we will emulate them.
> +    elif r.status == 405:
> +        h['can_zero'] = False
> +        h['can_trim'] = False
> +        h['can_flush'] = False
> +
> +    else:
> +        raise RuntimeError("could not use OPTIONS request: %d:
%s" %
> +                           (r.status, r.reason))
> +
> +def can_trim(h):
> +    get_options(h)
> +    return h['can_trim']
> +
> +def can_flush(h):
> +    get_options(h)
> +    return h['can_flush']
> +
> +def get_size(h):
> +    return params['disk_size']
> +
> +# For documentation see:
> +# https://github.com/oVirt/ovirt-imageio/blob/master/docs/random-io.md
> +# For examples of working code to read/write from the server, see:
> +#
>
https://github.com/oVirt/ovirt-imageio/blob/master/daemon/test/server_test.py
> +
> +def pread(h, count, offset):
> +    http = h['http']
> +    transfer=h['transfer']
> +    transfer_service=h['transfer_service']
> +
> +    http.putrequest("GET", h['path'])
> +    http.putheader("Authorization", transfer.signed_ticket)
>
This is useful only when working with old proxy that does not support PATCH.
Since this is pretty big (around 4k), better use it only when you use the
proxy
and h["can_zero"] is False.

The daemon does not know anything about Athorization header and new
proxy ignores it.

> +    http.putheader("Range", "bytes=%d-%d" % (offset,
offset+count-1))
> +    http.endheaders()
> +
> +    r = http.getresponse()
> +    # 206 = HTTP Partial Content, which is the usual response.
> +    if r.status != 200 and r.status != 206:
>
We never return 200 for Range request. Some server may return
200 but in this case they will also return the complete resource which
is completely useless in this context and will just corrupt the upload.

> +        h['transfer_service'].pause()
> +        h['failed'] = True
> +        raise RuntimeError("could not read sector (%d, %d): %d:
%s" %
> +                           (offset, count, r.status, r.reason))
> +    return r.read()
> +
> +def pwrite(h, buf, offset):
> +    count = len(buf)
> +    h['highestwrite'] = max(h['highestwrite'],
offset+count)
> +    real_pwrite(h, buf, offset, count)
> +
> +def real_pwrite(h, buf, offset, count):
> +    http = h['http']
> +    transfer=h['transfer']
> +    transfer_service=h['transfer_service']
> +
> +    http.putrequest("PUT", h['path'])
>
This will always flush - if you want to avoid flushing (you try to avoid in
zero), you need to specify:

    h["path"] + "?flush=n"

(not implemented yet but will be)

> +    http.putheader("Authorization", transfer.signed_ticket)
> +    # The oVirt server only uses the first part of the range, and the
> +    # content-length.
> +    http.putheader("Content-Range", "bytes %d-%d/*" %
(offset,
> offset+count-1))
> +    http.putheader("Content-Length", str(count))
> +    http.endheaders()
> +    http.send(buf)
>
What is the size of the typical buffers? sending new request for every
small buffer is not efficient.

> +
> +    r = http.getresponse()
> +    if r.status != 200:
> +        transfer_service.pause()
> +        h['failed'] = True
> +        raise RuntimeError("could not write sector (%d, %d): %d:
%s" %
> +                           (offset, count, r.status, r.reason))
+
> +def zero(h, count, offset, may_trim):
>
What is may_trim? trim can never replace zero.

> +    http = h['http']
> +    transfer=h['transfer']
> +    transfer_service=h['transfer_service']
> +
> +    # Unlike the trim and flush calls, there is no 'can_zero'
method
>
Do you mean h["can_zero"] set in getoptions()?

> +    # so nbdkit could call this even if the server doesn't support
> +    # zeroing.  If this is the case we must emulate.
> +    if not h['can_zero']:
> +        emulate_zero_with_pwrite(h, count, offset)
> +        return
> +
> +    # Construct the JSON request for zeroing.
> +    buf = json.dumps({'op': "zero",
> +                      'offset': offset,
> +                      'size': count,
> +                      'flush': False})
>
Is this intended? in cover letter you worte that no flush or trim
are done. Does qemu-img flush at the end?

> +
> +    http.putrequest("PATCH", h['path'])
> +    http.putheader("Content-Type", "application/json")
> +    http.putheader("Authorization", transfer.signed_ticket)
> +    http.putheader("Content-Length", len(buf))
> +    http.endheaders()
> +    http.send(buf)
> +
> +    r = http.getresponse()
> +    if r.status != 200:
> +        transfer_service.pause()
> +        h['failed'] = True
> +        raise RuntimeError("could not zero sector (%d, %d): %d:
%s" %
> +                           (offset, count, r.status, r.reason))
+
> +# qemu-img convert starts by trying to zero/trim the whole device.
> +# Since we've just created a new disk it's safe to ignore these
> +# requests as long as they are smaller than the highest write seen.
> +# After that we must emulate them with writes.
>
Isn't this needed also for the real zero? In the first version it will not
be very efficient. We don't want to zero the entire image.

If qemu-img wants to trim, better call trim - it will be very fast because
it will
do nothing in the first version :-)

> +def emulate_zero_with_pwrite(h, count, offset):
> +    if offset+count < h['highestwrite']:
> +        # count could be very large, so split into chunks.
> +        while count > 0:
> +            n = min(count, 65536)
+            buf = bytearray(n)
> +            real_pwrite(h, buf, offset, n)
> +            offset += n
> +            count -= n
>
This is not very efficient way to emulate zero. I think this can be better:

    http.putrequest("PUT", h['path'] + "?flush=n")
    http.putheader("Content-Range", "bytes %d-%d/*" %
(offset,
offset+count-1))
    http.putheader("Content-Length", str(count))
    http.endheaders()

    buf = bytearray(128 * 1024)

    while count > len(buf):
        http.send(buf)
        count -= len(buf)

    http.send(buffer(buf, 0, count))

    # check response...

See https://github.com/nirs/http-bench

+
> +def trim(h, count, offset):
> +    http = h['http']
> +    transfer=h['transfer']
> +    transfer_service=h['transfer_service']
> +
> +    # Construct the JSON request for trimming.
> +    buf = json.dumps({'op': "trim",
> +                      'offset': offset,
> +                      'size': count,
> +                      'flush': False})
> +
> +    http.putrequest("PATCH", h['path'])
> +    http.putheader("Content-Type", "application/json")
> +    http.putheader("Authorization", transfer.signed_ticket)
> +    http.putheader("Content-Length", len(buf))
> +    http.endheaders()
> +    http.send(buf)
> +
> +    r = http.getresponse()
> +    if r.status != 200:
> +        transfer_service.pause()
> +        h['failed'] = True
> +        raise RuntimeError("could not trim sector (%d, %d): %d:
%s" %
> +                           (offset, count, r.status, r.reason))
> +
> +def flush(h):
> +    http = h['http']
> +    transfer=h['transfer']
> +    transfer_service=h['transfer_service']
> +
> +    # Construct the JSON request for flushing.  Note the offset
> +    # and count must be present but are ignored by imageio.
> +    buf = json.dumps({'op': "flush",
> +                      'offset': 0,
> +                      'size': 0,
> +                      'flush': True})
>
flush does not accept any arguments - it will not fail, but better not have
code that has no effect.

> +
> +    http.putrequest("PATCH", h['path'])
> +    http.putheader("Content-Type", "application/json")
> +    http.putheader("Authorization", transfer.signed_ticket)
> +    http.putheader("Content-Length", len(buf))
> +    http.endheaders()
> +    http.send(buf)
> +
> +    r = http.getresponse()
> +    if r.status != 200:
> +        transfer_service.pause()
> +        h['failed'] = True
> +        raise RuntimeError("could not flush: %d: %s" %
(r.status,
> r.reason))
> +
> +def close(h):
> +    http = h['http']
> +    connection = h['connection']
> +
> +    http.close()
>
If qemu-img does not flush at the end, we can ensure flush here.
Maybe keep h["dirty"] flag, setting to True after very
write/zero/trim,
and setting the False after flush?

Then we can do:

    if h["ditry"]:
        flush(h.)

> +
> +    # If we didn't fail, then finalize the transfer.
> +    if not h['failed']:
> +        disk = h['disk']
> +        transfer_service=h['transfer_service']
> +
> +        transfer_service.finalize()
> +
> +        # Wait until the transfer disk job is completed since
> +        # only then we can be sure the disk is unlocked.  As this
> +        # code is not very clear, what's happening is that we are
> +        # waiting for the transfer object to cease to exist, which
> +        # falls through to the exception case and then we can
> +        # continue.
> +        endt = time.time() + timeout
> +        try:
> +            while True:
> +                time.sleep(1)
> +                tmp = transfer_service.get()
> +                if time.time() > endt:
> +                    raise RuntimeError("timed out waiting for
transfer to
> finalize")
> +        except sdk.NotFoundError:
> +            pass
> +
> +        # Write the disk ID file.  Only do this on successful completion.
> +        with builtins.open(params['diskid_file'], 'w') as
fp:
> +            fp.write(disk.id)
> +
> +    # Otherwise if we did fail then we should delete the disk.
> +    else:
> +        disk_service = h['disk_service']
> +        disk_service.remove()
>
This should be done only if we created the disk - is this always the case?

Maybe to make ths more robust, set a h["owns_disk"] flag when creating
the disk, and delete only in this case?

Nir

> +
> +    connection.close()
> diff --git a/v2v/rhv-upload-precheck.py b/v2v/rhv-upload-precheck.py
> new file mode 100644
> index 000000000..e3fa65148
> --- /dev/null
> +++ b/v2v/rhv-upload-precheck.py
> @@ -0,0 +1,72 @@
> +# -*- python -*-
> +# oVirt or RHV pre-upload checks used by ‘virt-v2v -o rhv-upload’
> +# Copyright (C) 2018 Red Hat Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License along
> +# with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free&entry=gmail&source=g>
> Software Foundation, Inc.,
> +# 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
> +
> +import json
> +import logging
> +import ovirtsdk4 as sdk
> +import ovirtsdk4.types as types
> +import sys
> +import time
> +
> +from http.client import HTTPSConnection
> +from urllib.parse import urlparse
> +
> +# Parameters are passed in via a JSON doc from the OCaml code.
> +# Because this Python code ships embedded inside virt-v2v there
> +# is no formal API here.
> +params = None
> +
> +if len(sys.argv) != 2:
> +    raise RuntimeError("incorrect number of parameters")
> +
> +# Parameters are passed in via a JSON document.
> +with open(sys.argv[1], 'r') as fp:
> +    params = json.load(fp)
> +
> +# What is passed in is a password file, read the actual password.
> +with open(params['output_password'], 'r') as fp:
> +    output_password = fp.read()
> +output_password = output_password.rstrip()
> +
> +# Parse out the username from the output_conn URL.
> +parsed = urlparse(params['output_conn'])
> +username = parsed.username or "admin@internal"
> +
> +# Connect to the server.
> +connection = sdk.Connection(
> +    url = params['output_conn'],
> +    username = username,
> +    password = output_password,
> +    ca_file = params['rhv_cafile'],
> +    log = logging.getLogger(),
> +    insecure = params['insecure'],
> +)
> +
> +system_service = connection.system_service()
> +
> +# Find if a virtual machine already exists with that name.
> +vms_service = system_service.vms_service()
> +vms = vms_service.list(
> +    search = ("name=%s" % params['output_name']),
> +)
> +if len(vms) > 0:
> +    vm = vms[0]
> +    raise RuntimeError("VM already exists with name ‘%s’, id
‘%s’" %
> +                       (params['output_name'], vm.id))
> +
> +# Otherwise everything is OK, exit with no error.
> diff --git a/v2v/test-v2v-python-syntax.sh b/v2v/test-v2v-python-syntax.sh
> new file mode 100755
> index 000000000..11ecd1f84
> --- /dev/null
> +++ b/v2v/test-v2v-python-syntax.sh
> @@ -0,0 +1,27 @@
> +#!/bin/bash -
> +# libguestfs
> +# Copyright (C) 2018 Red Hat Inc.
> +#
> +# This program is free software; you can redistribute it and/or modify
> +# it under the terms of the GNU General Public License as published by
> +# the Free Software Foundation; either version 2 of the License, or
> +# (at your option) any later version.
> +#
> +# This program is distributed in the hope that it will be useful,
> +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
> +# GNU General Public License for more details.
> +#
> +# You should have received a copy of the GNU General Public License
> +# along with this program; if not, write to the Free
>
<https://maps.google.com/?q=write+to+the+Free&entry=gmail&source=g>
> Software
> +# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
> 02110-1301 USA.
> +
> +set -e
> +
> +$TEST_FUNCTIONS
> +skip_if_skipped
> +
> +# Checks the files are syntactically correct, but not very much else.
> +python3 -m py_compile rhv-upload-createvm.py
> +python3 -m py_compile rhv-upload-plugin.py
> +python3 -m py_compile rhv-upload-precheck.py
> diff --git a/v2v/virt-v2v.pod b/v2v/virt-v2v.pod
> index 412fa0579..bc6631c3c 100644
> --- a/v2v/virt-v2v.pod
> +++ b/v2v/virt-v2v.pod
> @@ -6,15 +6,18 @@ virt-v2v - Convert a guest to use KVM
>
>   virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest
>
> - virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
> -   -o rhv -os rhv.nfs:/export_domain --network ovirtmgmt
> -
>   virt-v2v -i libvirtxml guest-domain.xml -o local -os /var/tmp
>
>   virt-v2v -i disk disk.img -o local -os /var/tmp
>
>   virt-v2v -i disk disk.img -o glance
>
> + virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
> +   -o rhv-upload -oc https://ovirt-engine.example.com/ovirt-engine/api \
> +   -os ovirt-data -op /tmp/ovirt-admin-password \
> +   --rhv-cafile /tmp/ca.pem --rhv-direct \
> +   --network ovirtmgmt
> +
>   virt-v2v -ic qemu:///system qemu_guest --in-place
>
>  =head1 DESCRIPTION
> @@ -42,9 +45,9 @@ libguestfs E<ge> 1.28.
>   Xen ───▶│ -i libvirt ──▶ │            │     │  (default) │
>   ... ───▶│  (default) │   │            │ ──┐ └────────────┘
>           └────────────┘   │            │ ─┐└──────▶ -o glance
> - -i libvirtxml ─────────▶ │            │ ┐└─────────▶ -o rhv
> - -i vmx ────────────────▶ │            │ └──────────▶ -o vdsm
> -                          └────────────┘
> + -i libvirtxml ─────────▶ │            │ ┐├─────────▶ -o rhv
> + -i vmx ────────────────▶ │            │ │└─────────▶ -o vdsm
> +                          └────────────┘ └──────────▶ -o rhv-upload
>
>  Virt-v2v has a number of possible input and output modes, selected
>  using the I<-i> and I<-o> options.  Only one input and output
mode can
> @@ -103,20 +106,18 @@ For more information see L</INPUT FROM VMWARE
> VCENTER SERVER> below.
>  =head2 Convert from VMware to RHV/oVirt
>
>  This is the same as the previous example, except you want to send the
> -guest to a RHV-M Export Storage Domain which is located remotely
> -(over NFS) at C<rhv.nfs:/export_domain>.  If you are unclear about
> -the location of the Export Storage Domain you should check the
> -settings on your RHV-M management console.  Guest network
> +guest to a RHV Data Domain using the RHV REST API.  Guest network
>  interface(s) are connected to the target network called
C<ovirtmgmt>.
>
>   virt-v2v -ic vpx://vcenter.example.com/Datacenter/esxi vmware_guest \
> -   -o rhv -os rhv.nfs:/export_domain --network ovirtmgmt
> +   -o rhv-upload -oc https://ovirt-engine.example.com/ovirt-engine/api \
> +   -os ovirt-data -op /tmp/ovirt-admin-password \
> +   --rhv-cafile /tmp/ca.pem --rhv-direct \
> +   --network ovirtmgmt
>
>  In this case the host running virt-v2v acts as a B<conversion
server>.
>
> -Note that after conversion, the guest will appear in the RHV-M Export
> -Storage Domain, from where you will need to import it using the RHV-M
> -user interface.  (See L</OUTPUT TO RHV>).
> +For more information see L</OUTPUT TO RHV> below.
>
>  =head2 Convert from ESXi hypervisor over SSH to local libvirt
>
> @@ -509,6 +510,10 @@ written.
>
>  This is the same as I<-o rhv>.
>
> +=item B<-o> B<ovirt-upload>
> +
> +This is the same as I<-o rhv-upload>.
> +
>  =item B<-o> B<qemu>
>
>  Set the output method to I<qemu>.
> @@ -533,6 +538,16 @@ I<-os> parameter must also be used to specify
the
> location of the
>  Export Storage Domain.  Note this does not actually import the guest
>  into RHV.  You have to do that manually later using the UI.
>
> +See L</OUTPUT TO RHV (OLD METHOD)> below.
> +
> +=item B<-o> B<rhv-upload>
> +
> +Set the output method to I<rhv-upload>.
> +
> +The converted guest is written directly to a RHV Data Domain.
> +This is a faster method than I<-o rhv>, but requires oVirt
> +or RHV E<ge> 4.2.
> +
>  See L</OUTPUT TO RHV> below.
>
>  =item B<-o> B<vdsm>
> @@ -632,6 +647,32 @@ virt-v2v finishes.
>
>  This disables progress bars and other unnecessary output.
>
> +=item B<--rhv-cafile> F<ca.pem>
> +
> +For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, the
F<ca.pem> file
> +(Certificate Authority), copied from F</etc/pki/ovirt-engine/ca.pem>
> +on the oVirt engine.
> +
> +=item B<--rhv-cluster> C<CLUSTERNAME>
> +
> +For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, set the RHV
Cluster
> +Name.  If not given it uses C<Default>.
> +
> +=item B<--rhv-direct>
> +
> +For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, if this option
is given
> +then virt-v2v will attempt to directly upload the disk to the oVirt
> +node, otherwise it will proxy the upload through the oVirt engine.
> +Direct upload requires that you have network access to the oVirt
> +nodes.  Non-direct upload is slightly slower but should work in all
> +situations.
> +
> +=item B<--rhv-verifypeer>
> +
> +For I<-o rhv-upload> (L<OUTPUT TO RHV>) only, verify the
oVirt/RHV
> +server’s identity by checking the server‘s certificate against the
> +Certificate Authority.
> +
>  =item B<--root ask>
>
>  =item B<--root single>
> @@ -1875,6 +1916,62 @@ Define the final guest in libvirt:
>
>  =head1 OUTPUT TO RHV
>
> +This new method to upload guests to oVirt or RHV directly via the REST
> +API requires oVirt/RHV E<ge> 4.2.
> +
> +You need to specify I<-o rhv-upload> as well as the following extra
> +parameters:
> +
> +=over 4
> +
> +=item I<-oc>
C<https://ovirt-engine.example.com/ovirt-engine/api>
> +
> +The URL of the REST API which is usually the server name with
> +C</ovirt-engine/api> appended, but might be different if you
installed
> +oVirt Engine on a different path.
> +
> +You can optionally add a username and port number to the URL.  If the
> +username is not specified then virt-v2v defaults to using
> +C<admin@internal> which is the typical superuser account for oVirt
> +instances.
> +
> +=item I<-op> F<password-file>
> +
> +A file containing a password to be used when connecting to the oVirt
> +engine.  Note the file should contain the whole password, B<without
> +any trailing newline>, and for security the file should have mode
> +C<0600> so that others cannot read it.
> +
> +=item I<-os> C<ovirt-data>
> +
> +The storage domain.
> +
> +=item I<--rhv-cafile> F<ca.pem>
> +
> +The F<ca.pem> file (Certificate Authority), copied from
> +F</etc/pki/ovirt-engine/ca.pem> on the oVirt engine.
> +
> +=item B<--rhv-cluster> C<CLUSTERNAME>
> +
> +Set the RHV Cluster Name.  If not given it uses C<Default>.
> +
> +=item I<--rhv-direct>
> +
> +If this option is given then virt-v2v will attempt to directly upload
> +the disk to the oVirt node, otherwise it will proxy the upload through
> +the oVirt engine.  Direct upload requires that you have network access
> +to the oVirt nodes.  Non-direct upload is slightly slower but should
> +work in all situations.
> +
> +=item I<--rhv-verifypeer>
> +
> +Verify the oVirt/RHV server’s identity by checking the server‘s
> +certificate against the Certificate Authority.
> +
> +=back
> +
> +=head1 OUTPUT TO RHV (OLD METHOD)
> +
>  This section only applies to the I<-o rhv> output mode.  If you use
>  virt-v2v from the RHV-M user interface, then behind the scenes the
>  import is managed by VDSM using the I<-o vdsm> output mode (which
end
> --
> 2.13.2
>
> _______________________________________________
> Libguestfs mailing list
> Libguestfs@redhat.com
> https://www.redhat.com/mailman/listinfo/libguestfs