PROBLEM: (85583, CFS_81658) (PATCH ID: TCR510-027) ******** This patch fixes the following issues: - When the path to a tape or media_changer device is lost, CAA does not update the state of the corresponding CAA resources to offline. Consequently, application resources that depend on the disconnected tape or media_changer are not failed over. Without this patch, the workaround was to unregister the tape or media changer resource then re-register it. - When the path to a disconnected tape or media_changer device is restored, CAA does not update the state of the corresponding CAA resource to online. Consequently, application resources that depend on the restored tape or media_changer can not be started. Without this patch, the workaround was to unregister the tape or media changer resource then re-register it. - The problem was caused by a change in the event name generated when the path to the device is lost or restored. The new event names were ignored by CAA. - Under certain conditions an event needed by one resource monitor may be prematurely discarded by another monitor. This causes CAA to miss the event, thereby failing to update the resource state. In this case, the following message will be logged in /var/adm/messages/syslog.dated/current/daemon.log: CAAD[]: Invalid EVM Event passed in PROBLEM: (117-1-16757) (PATCH ID: TCR510-067) ******** This patch fixes a situation when CAA daemon on a clustered system crashes and dumps core. PROBLEM: (117-1-19150) (PATCH ID: TCR510-110) ******** This patch fixes the major problems of CAA applications not failing over during a node shutdown and caad hang condition at startup. PROBLEM: (91894, 90159, 90797, 77411) (PATCH ID: TCR510-121) ******** Problem: no-roll patch cluster_lockd is not started. After no-roll patch, and before the noroll_versw command was run, the following information is seen from the console: CAA daemon started NFS mount daemon started NFS export service started Dependency analysis failed because of: Resource in UNKNOWN state: cluster_lockd Resource cluster_lockd has placement error. NFS IO service started Output from daemon.log this morning prior to cluster halt and reboot after running noroll_versw command. root@brack:/usr/bin > Feb 8 08:32:12 brack CAAD[524807]: `cluster_lockd` on `brack` went OFFLINE unexpectedly Feb 8 08:32:12 brack CAAD[524807]: Attempting to stop `cluster_lockd` on member`brack` Feb 8 08:32:14 brack CAAD[524807]: `cluster_lockd` on member `brack` has experienced an unrecoverable failure. Feb 8 08:32:14 brack CAAD[524807]: Human intervention required to resume its availability. Feb 8 08:32:14 brack CAAD[524807]: `cluster_lockd` experienced a failure on `brack`. Stopping dependent resources. Feb 8 08:32:14 brack CAAD[524807]: Attempting to stop `cluster_lockd` on member `brack` Feb 8 08:32:16 brack CAAD[524807]: `cluster_lockd` on member `brack` has experienced an unrecoverable failure. Problem: Caa resources are not started. Create a caa resource and register it. Then try to start it using caa_start. Caa_start will return an error message indicating that the resource could not be started. Then look at the resource using caa_stat. The resource will be in the unknow state. There will be a core file of the caa daemon, however /usr/sbin/caad will still be running. # caa_profile -create App1 -t application -B \ /usr/test/chart/components/caa/bin/DaemonApplication1 # echo $? 0 # caa_register App1 # echo $? 0 # caa_start App1 Attempting to start `App1` on member `ackbar` `App1` on member `ackbar` has experienced an unrecoverable failure. Human intervention required to resume its availability. Could not start resource App1. # ls -l /var/cluster/caa/ackbar* total 27448 -rw------- 1 root system 5521408 Oct 3 16:07 core.caad.ackbar.0 -rw------- 1 root system 5496832 Oct 3 16:07 core.caad.ackbar.1 # caa_stat -t App1 Name Type Target State Host ------------------------------------------------------------ App1 application ONLINE UNKNOWN ackbar