FibreChannel TOV

[Marco A.]

Hi,

what's the recommended Time out value for a cache centric arrays, I was requested to set up as 60 sec, it is recommended .

If we are running under a clustered env , it is going to affect us ?

Regards,

Marc'0

[IBaltay · Accepted solution]

Hi,
there are special recommendations to change the TOV especially when we have the EVA connected to the XP (e.g. EVA 8100 to XP 24000) as an external storage - for the online EVA 8100 firmware upgrade procedure
both IO_TOV and PBWT should be set to 180 seconds prior to the EVA controller firmware upgrade with the reset to the defaults after the procedure (the defaults are the following:
IO_TOV=15 adn PBWT=10 seconds.
Can you provide the cluster/application specification and the reason for the 60s IO_TOV recommendation pls?

[Eric de Lange (MSE)]

Marco,

In general HP recommend that the SAN hosts uses a >SCSI< command timeout of 60 when e.g. connected to an XP. Depending on the size of the SAN environment and the configuration sometimes 120 seconds is more appropriate.

If you are running in a clustered environment this means that the cluster node will have to wait for this timeout period before it can take corrective measures. Usually this will force a path failover to the alternate path. If even that fails the cluster could decide to failover the package to the alternate clusternode.

60 seconds a normal value and is recommended in various XP Configuration guides.

hope this helps.

-Eric