FAS3250 NFS performance degradation with a full aggregate
I have a FAS3250 in my environment we’re keeping around for NFS storage on old HPUX/IRIX and SunOS systems. Our users have been complaining about this NFS storage being unusably slow.
I’ve been driving myself crazy trying to figure out what the issue is. I used nfsstat to show clients consuming the most network bandwidth, went to those hosts, and there are no active processes reading or writing to the FAS3250. At first a couple Solaris systems were showing as the top clients, I had the users pause doing their software stuff on them which took them off the list, but our RHEL7 server we use as a gateway between our FAS3250 and FAS70 to do backups will move up the list to be at the top and I’ve confirmed there were no backups running. (xcp as a cronjob)
I did a read/write test on the FAS3250 from the same RHEL7 system, it’s slow, about 50 mb/s.
We do not touch the FAS3250 unless something is wrong with it, no one on our team is comfortable messing with it and naturally support for it now is non-existent.
Network hardware hasn’t changed. The system is using a single 10GBe port for data.
Rambling aside, my question is, the NFS volumes are on a 96% full aggregate. Would this cause the performance issues I’m seeing? The volumes themselves have a good amount of space left. I’m guessing they were thick provisioned.
I’m not great with storage or networking if you couldn’t tell. If anyone has any insight I’d really appreciate it.