topic Re: Polyserve failover in Storage Software

Polyserve failover

jknightcs1 — Thu, 26 Feb 2009 20:29:52 GMT

We've been using a two-node PolyServe Matrix Server cluster for NFS/Cifs hosting for a while now, on two linux hosts. Recently, we noticed that during some failovers of virtual IP's servicing NFS between nodes, some client systems would fail to see the NFS share for 1-5 minutes, as if they couldn't see the new location of the IP. Not the same system every time, and not always for the same amount of time, but it's happened twice so far.
Anyone else seen this? More importantly, how does the failover work? Is it just a gratuitous ARP like Heartbeat?

Re: Polyserve failover

Emil Velez — Sat, 28 Feb 2009 01:03:31 GMT

Depends on NFS client. NFS client has to timeout and the re-arp the IP address. So this is somehwhat totally depenent on the client and the timeout of ARP information in the cache.

Re: Polyserve failover

jknightcs1 — Mon, 02 Mar 2009 19:48:35 GMT

Any way to change that setting?

Re: Polyserve failover

Emil Velez — Sun, 22 Mar 2009 23:38:38 GMT

You would have to go into your clients and see if you can change the re-arp interval.