Libguestfs - Oct 2017 - [PATCH] daemon: proto: Remove pervasive but useless debugging messages.

If you spend your time looking at libguestfs debugging output you'll
see many messages from the daemon main loop like this:

  guestfsd: main_loop: new request, len 0x54
  guestfsd: main_loop: proc 278 (mkfs) took 0.02 seconds

I don't think these messages really bring much value.  This commit
removes them entirely.

An alternative might be to change them to make them shorter and/or
less useless, but I'm not sure what that would be.

Rich.

---
 daemon/proto.c | 25 -------------------------
 1 file changed, 25 deletions(-)

diff --git a/daemon/proto.c b/daemon/proto.c
index 14f7efe94..670ae2ffb 100644
--- a/daemon/proto.c
+++ b/daemon/proto.c
@@ -96,11 +96,6 @@ main_loop (int _sock)
     xdr_u_int (&xdr, &len);
     xdr_destroy (&xdr);
 
-    if (verbose)
-      fprintf (stderr,
-	       "guestfsd: main_loop: new request, len 0x%" PRIx32
"\n",
-	       len);
-
     /* Cancellation sent from the library and received after the
      * previous request has finished processing.  Just ignore it.
      */
@@ -188,26 +183,6 @@ main_loop (int _sock)
     /* Now start to process this message. */
     dispatch_incoming_message (&xdr);
     /* Note that dispatch_incoming_message will also send a reply. */
-
-    /* In verbose mode, display the time taken to run each command. */
-    if (verbose) {
-      struct timeval end_t;
-      gettimeofday (&end_t, NULL);
-
-      int64_t start_us, end_us, elapsed_us;
-      start_us = (int64_t) start_t.tv_sec * 1000000 + start_t.tv_usec;
-      end_us = (int64_t) end_t.tv_sec * 1000000 + end_t.tv_usec;
-      elapsed_us = end_us - start_us;
-
-      fprintf (stderr,
-	       "guestfsd: main_loop: proc %d (%s) took %d.%02d seconds\n",
-               proc_nr,
-               proc_nr >= 0 && proc_nr <= GUESTFS_MAX_PROC_NR
-               ? function_names[proc_nr] : "UNKNOWN PROCEDURE",
-               (int) (elapsed_us / 1000000),
-               (int) ((elapsed_us / 10000) % 100));
-    }
-
   cont:
     xdr_destroy (&xdr);
     free (buf);
-- 
2.13.2

On Monday, 9 October 2017 18:20:53 CEST Richard W.M. Jones
wrote:
> If you spend your time looking at libguestfs debugging output you'll
> see many messages from the daemon main loop like this:
> 
>   guestfsd: main_loop: new request, len 0x54
>   guestfsd: main_loop: proc 278 (mkfs) took 0.02 seconds
> 
> I don't think these messages really bring much value.  This commit
> removes them entirely.
TBH these information were useful to me in the past:
a) the request length made me discover that a locally generated file,
   uploaded to the guest with copy/copy_in, was way too bigger than
   it was supposted to me
b) the timing information is actually helpful to diagnose slow API
   calls (hello, vmware-uninstall-tools.pl)
> An alternative might be to change them to make them shorter and/or
> less useless, but I'm not sure what that would be.
Unfortunately I don't have alternative solution, I'm afraid.  The only
bit that I never used was the proc number, since its name is way more
helpful.

-- 
Pino Toscano