I'm periodically seeing NFS server timeouts in my messages and amq logs shortly after a "service osad restart" is run via cron. Once the NFS server times out it fails over to my passive node. My mtab values for the nfs mount are set to timeo=10,retrans=2. Wondering if these values are too low and if the NFS hiccup could be caused by the cron? Specifically its running:
/sbin/service osad restart >/dev/null 2>&1 ; /usr/sbin/rhn_check >/dev/null 2>&1 ; /usr/sbin/rhn-profile-sync

Thanks for any information anyone can share.

psag_linux

I think the default timeo is 600 for tcp.
Why don't you leave the default?

I've recently taken over support for the servers, not sure why they went with timeo=10 which if I understand it correctly will give me 2 seconds to respond to NFS with 2 reties.

The unit is tenth of a second; timeo=10 means 1 second.
This includes the time to establish a TCP connection, and is too short in case the NFS server is slow/busy.
As TCP has its own retry meachanism (e.g. every 10 seconds), a multiple of it makes sense for timeo (200,300,400,...).

Thanks for the input.