Poor IO performance after RAID failure

From: Hallen, Mark (Mark.Hallen@savvis.net)
Date: Tue Aug 24 2004 - 14:48:28 EDT


Hello,

We have a Sun-Fire-480R running Solaris 8 which has a GForce 2400 RAID
unit connected to an Antares P-0076 Ultra 160 HBA. We general see about
65M/s for write performance. When the RAID fails for some reason (I am
powering it down for testing) and the system doesn't get rebooted, the
write performance goes down to about 6M/s. We are trying to implement
database mirroring across 2 of these RAIDs and it would be extremely
nice to be able to maintain the original write performance when
something fails without having to reboot the system. I have tried the
following:

umount the fs

cfgadm -c unconfigure c3

cfgadm -c configure c3

mount the fs

I would think that this would reset and probe the bus and we would be
good to go, but I still see only 6M/s or so until the system is
rebooted. I don't see any errors in the system log.

Any suggestions or info would be greatly appreciated. I searched
sunsolve and didn't come up with anything applicable.

Thanks much,

Mark Hallen
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers



This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:29:20 EDT