Red Hat Knowledgebase

Release Found:Red Hat Enterprise Linux 4

Description:

There is a known issue in clvmd where the daemon does not properly update its cache after the underlying storage of a volume group is changed.

For example, a new physical volume is created, and then it is added to a volume group using the vgextend command. From this point forward, all LVM commands fail with errors similar to the following:

Error locking on node NODE01: Internal lvm error, check syslog  
Error locking on node NODE02: Internal lvm error, check syslog
Error locking on node NODE03: Internal lvm error, check syslog

In the syslog, messages similar to these appear:

Nov  8 16:31:36 NODE01 lvm[4099]: Couldn't find device with uuid
'zNi9rm-cc3K-nymr-seyM-pWp7-FaLT-Uz0JbP'.
Nov  8 16:31:36 NODE02 lvm[4099]: Couldn't find all physical volumes
for volume group joe.
Nov  8 16:31:36 NODE03 lvm[4099]: Couldn't find device with uuid
'zNi9rm-cc3K-nymr-seyM-pWp7-FaLT-Uz0JbP'.

Workaround

For versions of lvm2-cluster-2.02.21-7 and later, a new switch has been added to clvmd. The issue can be avoided by running the following command after the underlying storage is changed:

clvmd -R

This command should be run on all nodes.

For versions earlier than lvm2-cluster-2.02.21-7, it is necessary to kill and restart the clvmd processes on all nodes. When the clvmd process restart, the cache is cleared.

For more information, see: https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=138396