Services using HA-LVM with no "lv_name" attribute are not recovered after a disk failure in RHEL 5 or 6
Issue
- Services using HA-LVM are not relocated after disk access failure.
- We've found that, if a node loses access to the storage,
lvmresource can not deactivate the VG for the services running in the node: - When I simulate a disk failure scenario,
lvmresource recovery does not always work as expected. - After all physical volumes in a volume group managed by an
lvmresource fail, the other node cannot recover it:
Oct 1 18:07:17 node2 rgmanager[10988]: [lvm] Starting volume group, myvg
Oct 1 18:07:17 node2 rgmanager[11027]: [lvm] Someone else owns this volume group
Environment
- Red Hat Enterprise Linux (RHEL) 5 or 6 with the High Availability Add On
- HA-LVM (either
clvmdortaggingbased)- Using a "by-vg" configuration: One or more
lvmresources in/etc/cluster/cluster.confthat do not have anlv_nameattribute specified
- Using a "by-vg" configuration: One or more
- RHEL 6:
resource-agentsprior to release3.9.2-12.el6_3.2or release3.9.2-21.el6 - RHEL 5:
rgmanagerprior to release2.0.52-37.el5_9.4or releasergmanager-2.0.52-47.el5
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase, tools, and much more.