HP XP P9500 Storage - Manual (Page 17)

Figure 4 Failover after primary system failure
ESAM also performs the following checks to detect failures:
•
The RCU issues service information messages (SIMs) when the data path is blocked. The
multipath software issues messages about the failure in the host-MCU paths.
•
Health check of the quorum disk by the MCU and RCU. The primary or secondary system
issues a SIM if a failure in the quorum disk is detected. Host operations will not switch to the
S-VOL if the quorum disk fails. In this case, the failure must be cleared as soon as possible
and the quorum disk recovered.
•
If the multipath software detects a failure in the host-to-pair volume paths, the operation switches
to a different available path and no SIM is issued. To stay informed about path status, monitor
the path failure messages issued by the multipath software.
•
The multipath software issues message(s) when all host-MCU paths fail. These messages must
then be checked and the cause corrected. If failover took place, host operations should be
switched back to the primary system.
It is possible that maintenance operations require both storage systems to be powered off at the
same time. In this case, the health checking periods would be shortened to prevent unexpected
failover while both systems are powered off. See
“Planned outage of primary and secondary
systems, quorum disk” (page 46) for more information.
After failover, when a failure is corrected, you may continue operations on the S-VOL, though HP
recommends switching them back to the P-VOL. To find which volume was originally a P-VOL, use
the multipath software on the host to refer to the path information, checking for the volume with
the owner path. The owner path is set to the volume that you specified as a P-VOL when you created
an ESAM pair. The owner path never switches even if the P-VOL and S-VOL were swapped due
to a failover.
Preventing unnecessary failover
Some applications issue the read command to the ESAM S-VOL. When these applications are
used, and when the number of read commands to the S-VOL reaches or exceeds the threshold
(1,000 times per six minutes), ESAM assumes that a P-VOL failure has occurred. This situation
results in an unnecessary failover to the ESAM S-VOL.
At this time, the Solaris VERITAS Volume Manager (VxVM) vxdisksetup command issues more
read commands than allowed by the threshold.
You can prevent unnecessary failover by setting host mode option 48 to ON. Note that when this
option is ON, the S-VOL responds slower to the read command.
Planning failover 17