Red Hat Bugzilla – Bug 135730
init.d/lock_gulmd should not stop if GFS is mounted
Last modified: 2010-01-11 21:59:49 EST
Description of problem:
The init.d init script should probably not stop if there are resources
running that require it (i.e. GFS and/or GNBD). Stopping lock_gulmd
on a client mounting GFS can cause orphaned locks to appear in the
lock server (see bug# 129193). Additional logic that checked to see
if the resource was in use would help prevent the occurance of this bug.
I would like to make the stop operation fail in the event that GFS is
currently mounted on the system. Operaters running this script
manually will then be informed that the lock_gulmd service is unable
to stop because it is already in use.
The current stop operation will become a "forcestop" operation so that
this functionality is not lost.
Version-Release number of selected component (if applicable):
GFS-6.0.0-15 and earlier
The cleanest way that I can think to do this requires accessing the
Gulm_Service information of the local lock_gulmd core on the node
calling init.d/lock_gulmd stop. This information is currently only
available via the SIGUSR1 debugging dumps. A cleaner resource
requesting method has been requested (bug #136220).
this has been resolved in version 126.96.36.199 of gulm/init.d/lock_gulmd
An errata has been issued which should help the problem
described in this bug report. This report is therefore being
closed with a resolution of ERRATA. For more information
on the solution and/or where to find the updated files,
please follow the link below. You may reopen this bug report
if the solution does not work for you.