Bug 190392

Summary: cman init scripts seem to always fail when stopping
Product: [Retired] Red Hat Cluster Suite Reporter: Corey Marthaler <cmarthal>
Component: cmanAssignee: Abhijith Das <adas>
Status: CLOSED ERRATA QA Contact: Cluster QE <mspqa-list>
Severity: medium Docs Contact:
Priority: medium    
Version: 4CC: cfeist, cluster-maint
Target Milestone: ---   
Target Release: ---   
Hardware: All   
OS: Linux   
Whiteboard:
Fixed In Version: RHBA-2006-0556 Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2006-08-10 21:32:59 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
fix for bz none

Description Corey Marthaler 2006-05-01 21:32:53 UTC 
Description of problem:
I've been seeing this a lot lately and it's annoying me. :)

I'll try and debug this a bit more later but here's what I've been seeing:

[root@taft-01 ~]# cat /proc/cluster/services
Service          Name                              GID LID State     Code
[root@taft-01 ~]# service cman stop
Stopping cman:                                             [FAILED]
[root@taft-01 ~]# cat /proc/cluster/nodes
Node  Votes Exp Sts  Name

from syslog:
May  1 10:30:42 taft-01 kernel: CMAN: we are leaving the cluster.
May  1 10:30:42 taft-01 ccsd[8725]: Cluster manager shutdown.  Attemping to
reconnect...
May  1 10:30:45 taft-01 cman: failed to stop cman failed

Version-Release number of selected component (if applicable):
[root@taft-01 ~]# rpm -q cman
cman-1.0.4-0

How reproducible:
almost every time
Comment 1 Christine Caulfield 2006-05-02 07:04:06 UTC 
Needs passing to whever manages the init scripts. All the subsystems need to be
shut down. This includes ccsd which polls cman to see if it is active.
Comment 2 Corey Marthaler 2006-05-02 14:55:44 UTC 
First, when did that change? Ccsd must be started before cman can be started and
we've always (since the begining of rhel4) stopped cman before stopping ccsd and
it worked just fine.

Second, stopping ccsd first still doen't help:
[root@taft-02 ~]# service ccsd stop
Stopping ccsd:                                             [  OK  ]
[root@taft-02 ~]# service cman stop
Stopping cman:                                             [FAILED]

In both cases (before and after ccsd is stopped), a leave by hand shows that the
stop did infact work:
[root@taft-02 ~]# cman_tool leave
cman_tool: Error leaving cluster: Cluster software not started
Comment 4 Abhijith Das 2006-05-02 19:28:53 UTC 
Created attachment 128512 [details]
fix for bz

The cman init script ties the success of 'service cman stop' to successfully
completing a 'modprobe -r cman'. The 'modprobe -r' doesn't always succeed
because other modules (gfs, lock_dlm) may still be using it. This patch ties
the success of 'service cman stop' to 'cman_tool leave' which makes more sense
to me. Let me know if this works for you and I'll commit the patch.
Comment 5 Corey Marthaler 2006-05-08 21:46:57 UTC 
Having the success tied to cluster membership removal does make more sense. The
patch works for me.
Comment 6 Abhijith Das 2006-05-08 22:32:09 UTC 
Checked in fix to RHEL4, HEAD and STABLE.
Comment 7 Corey Marthaler 2006-08-04 16:09:28 UTC 
fix verified.
Comment 9 Red Hat Bugzilla 2006-08-10 21:32:59 UTC 
An advisory 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.

http://rhn.redhat.com/errata/RHBA-2006-0556.html