|Summary:||RFE: add the ability to get a lockdump out of gfs through sysrq|
|Product:||[Retired] Red Hat Cluster Suite||Reporter:||Josef Bacik <jbacik>|
|Component:||GFS-kernel||Assignee:||Robert Peterson <rpeterso>|
|Status:||CLOSED WONTFIX||QA Contact:||Cluster QE <mspqa-list>|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||2008-03-25 17:38:50 UTC||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
|Bug Depends On:|
|Bug Blocks:||295531, 295541|
Description Josef Bacik 2007-07-31 21:27:38 UTC
As more customers are starting to use GFS in a shared root environment its becoming more difficult to troubleshoot gfs hangs, as we cannot run gfs_tool lockdump because a GFS hang usually means the entire box is hung and the only way to get information is via sysrq. This patch dumps glock information via sysrq. I've tested this patch on my cluster and doesn't appear to blow anything up.
Comment 1 Josef Bacik 2007-07-31 21:27:38 UTC
Created attachment 160362 [details] patch that implements this functionality.
Comment 3 Robert Peterson 2008-03-25 17:38:50 UTC
While I still feel this would be useful, it requires registering a new sysrq sequence. Since the sysrq sequences are designed and implemented by the upstream community, and gfs1 is not upstream (nor will it ever be) the decision was made to not implement this fix. It would be too easy for upstream to make changes that are not compatible with the fix. So let's keep it a patch for now.