Joe Little wrote: > > I will confess to not knowing how to measure latency. I haven't > observed the issue with UFS, but I haven't been trying much since UFS > doesn't interest me as much. That said, our UFS only Solaris 9/10 > machines haven't had this issue. > Hi Joe,
I've attached a very basic dtrace script to measure latency for all NFS3 server ops. Notice that this script measures the latency incurred both from NFS3 server processing and the underlying file system. Just run it as root on your server and crank up a NFS3 client. It will produce the following metrics: 1. op count 2. avg response time per op 3. avg system time consumed per op 4. quantized avg resp time 5. quantized avg system time I recommend running the same workload against a NFS mounted tmpfs (to set the utopian baseline), UFS, and ZFS. Then look for the large differences. That will give us an idea of which NFS3 procs are having perf issues, and we can dig deeper. I'm out of town Friday and Monday, but Spencer will cover for me if you run into any problems. Jeff -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: srv3snoop.d URL: <http://mail.opensolaris.org/pipermail/nfs-discuss/attachments/20060504/02a7e90f/attachment.ksh>
