Skip to content

Forward port of RHEL6 bug fixes #150

Merged
merged 9 commits into from Oct 12, 2012

5 participants

@ryan-mccabe

These commits forward port a handful of bugs that were fixed for RHEL6

ryan-mccabe added some commits Oct 8, 2012
@ryan-mccabe ryan-mccabe resource-agents: Fix return code in fslib.sh
Patch from John Ruemker <jruemker@redhat.com>:

When you have an fs resource that depends on an lvm resource in a service,
and that lvm resource fails to start, the subsequent attempt to stop the fs
resource will fail because the device doesn't exist (since the logical
volume was never activated).

Because we're stopping a resource that is not actually running
(the fs isn't mounted), the operation should succeed here. Instead it
fails, preventing recovery.

Resolves: rhbz#853249
Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
1433f09
@ryan-mccabe ryan-mccabe resource-agents: Fix for reading /proc/mounts in fs-lib.sh
This patch fixes a problem with the fs resource agent
that caused status checks to mistakenly fail as a result
of /proc/mounts changing while it was being read.

Original patch from John Ruemker <jruemker@redhat.com>

Resolves rhbz#853249

Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
6632940
@ryan-mccabe ryan-mccabe resource-agents: Add debugging output to fs-lib.sh
Add debugging output to the status/monitor operation in
fs-lib.sh analagous to what other agents do.

Resolves: rhbz#714156

Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
8a86560
@ryan-mccabe ryan-mccabe resource-agents: Set restart retries to 0 for oracledb.sh
Per the suggestion of Chris Phillips <cphillip@redhat.com> change
the setting of RESTART_RETRIES to 0 from 3 to prevent rgmanager from
attempting to restart the database during the status operation so that
failures are properly counted.

Resolves: rhbz#817550

Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
2a3b233
@ryan-mccabe ryan-mccabe resource-agents: Fix tomcat-6 startup
The /usr/sbin/tomcat6 script no longer parses the configuration file
and sets shell variables before trying to start tomcat6. Do that manually
so that we can actually run.

Resolves: rhbz#757632
Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
a7e24ea
@fghaas
Cluster Labs member
fghaas commented Oct 9, 2012

I'm afraid I'm a bit underqualified to properly review RAs in the rgmanager OCF flavor. @fabbione or @lhh, can you help?

@fabbione
Lon Hohberger and others added some commits Oct 10, 2012
Lon Hohberger Med: Allow LVM return codes to work with Pacemaker
rgmanager treated any non-zero code as an error.  Unfortunately,
Pacemaker relies on correct OCF_NOT_RUNNING return codes in the
case that a resource is cleanly stopped.

Acked-by: Ryan McCabe <rmccabe@redhat.com>
Acked-by: Jonathan Brassow <jbrassow@redhat.com>
Signed-off-by: Lon Hohberger <lhh@redhat.com>
d19b7d2
Lon Hohberger Med: Update member_util to use corosync and pacemaker
The LVM agent was using rgmanager tools to determine
ownership of LVs; this is fine except when rgmanager
doest not exist on the system.

Acked-by: Ryan McCabe <rmccabe@redhat.com>
Signed-off-by: Lon Hohberger <lhh@redhat.com>
f068c04
@ryan-mccabe ryan-mccabe Med: Strip whitespace before checking for empty string in HA LVM
Using vgs to check for a tag with 'vgs -o tags --noheadings' will
produce 9 space characters when no tags are found. This patch strips
that space before checking whether the output is empty.

Patch originally from Frank Danapfel <fdanapfe@redhat.com>

Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
6f3151b
@ryan-mccabe ryan-mccabe Med: More cleanups for whitespace in output in HA LVM
Clean up a few more places where we call lvs or vgs -o tags
to strip whitespace before we check for blank output.

Reviewed-by: Lon Hohberger <lhh@redhat.com>
Reviewed-by: Frank Danapfel <fdanapfe@redhat.com>
Signed-off-by: Ryan McCabe <rmccabe@redhat.com>
585b3dd
@lhh
lhh commented Oct 11, 2012

Note that Jon Brassow already reviewed the changes to lvm*.sh. Also, they were tested (including the member_util.sh change) with Pacemaker 1.1.8 and rgmanager-3.0.12.1

@feist feist merged commit 5f082fb into ClusterLabs:master Oct 12, 2012
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Something went wrong with that request. Please try again.