Ovirt devel - Mar 2010 - [PATCH node] Fix hang when cleaning dirty storage

Code that scanned for dirty storage was accidentally grepping the
contents of an entire partition.  This led to an apparent hang in
situations where partitions were large.

Signed-off-by: Mike Burns <mburns at redhat.com>
---
 scripts/ovirt-config-storage |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/scripts/ovirt-config-storage b/scripts/ovirt-config-storage
index aea73f8..92d79dd 100755
--- a/scripts/ovirt-config-storage
+++ b/scripts/ovirt-config-storage
@@ -550,7 +550,7 @@ wipe_lvm_on_disk()
     unmount_logging
     for vg in $(pvs -o vg_name --noheadings $HOSTVGDRIVE* 2>/dev/null|sort
-u); do
         if pvs -o pv_name,vg_name --noheadings | \
-            grep $vg | grep -v -q $dev* 2>/dev/null; then
+            grep $vg | egrep -v -q "${dev}p?[0-9]*" 2>/dev/null;
then
             log "The volume group \"$vg\" spans multiple
disks."
             log "This operation cannot complete.  Please manullay"
             log "cleanup the storage using standard linux tools."
-- 
1.6.6.1

Patch has problems is there are a large number of devices.  A re-worked
patch should follow shortly.

Mike

On Thu, 2010-03-11 at 14:55 -0500, Mike Burns wrote:
> Code that scanned for dirty storage was accidentally grepping the
> contents of an entire partition.  This led to an apparent hang in
> situations where partitions were large.
> 
> Signed-off-by: Mike Burns <mburns at redhat.com>
> ---
>  scripts/ovirt-config-storage |    2 +-
>  1 files changed, 1 insertions(+), 1 deletions(-)
> 
> diff --git a/scripts/ovirt-config-storage b/scripts/ovirt-config-storage
> index aea73f8..92d79dd 100755
> --- a/scripts/ovirt-config-storage
> +++ b/scripts/ovirt-config-storage
> @@ -550,7 +550,7 @@ wipe_lvm_on_disk()
>      unmount_logging
>      for vg in $(pvs -o vg_name --noheadings $HOSTVGDRIVE*
2>/dev/null|sort -u); do
>          if pvs -o pv_name,vg_name --noheadings | \
> -            grep $vg | grep -v -q $dev* 2>/dev/null; then
> +            grep $vg | egrep -v -q "${dev}p?[0-9]*"
2>/dev/null; then
>              log "The volume group \"$vg\" spans multiple
disks."
>              log "This operation cannot complete.  Please
manullay"
>              log "cleanup the storage using standard linux
tools."

Code that scanned for dirty storage was accidentally grepping the
contents of an entire partition.  This led to an apparent hang in
situations where partitions were large.

Signed-off-by: Mike Burns <mburns at redhat.com>
---
 scripts/ovirt-config-storage |    8 ++++++--
 1 files changed, 6 insertions(+), 2 deletions(-)

diff --git a/scripts/ovirt-config-storage b/scripts/ovirt-config-storage
index aea73f8..1e81db0 100755
--- a/scripts/ovirt-config-storage
+++ b/scripts/ovirt-config-storage
@@ -548,9 +548,13 @@ wipe_lvm_on_disk()
 {
     local dev=${1-$HOSTVGDRIVE}
     unmount_logging
-    for vg in $(pvs -o vg_name --noheadings $HOSTVGDRIVE* 2>/dev/null|sort
-u); do
+    local part_delim="p"
+    if [[ "$dev" =~ "/dev/sd" ]]; then
+        part_delim=""
+    fi
+    for vg in $(pvs -o vg_name --noheadings $HOSTVGDRIVE
$HOSTVGDRIVE${dev_delim}[0-9]* 2>/dev/null|sort -u); do
         if pvs -o pv_name,vg_name --noheadings | \
-            grep $vg | grep -v -q $dev* 2>/dev/null; then
+            grep $vg | egrep -v -q "${dev}${part_delim}[0-9]+|${dev}"
2>/dev/null; then
             log "The volume group \"$vg\" spans multiple
disks."
             log "This operation cannot complete.  Please manullay"
             log "cleanup the storage using standard linux tools."
-- 
1.6.6.1