Home / server / Questions / 280702
In Process
NPE
Asked: 2011-06-16 07:01:02 +0800 CST

NFS performance troubleshooting

  • 772

I am troubleshooting NFS performance issues on Linux, and I'm looking at the following nfsiostat output:

host:/path mounted on /path:

   op/s   rpc bklog
  96.75      0.01
read:       ops/s        kB/s      kB/op    retrans    avg RTT (ms)    avg exe (ms)
           86.561    1408.294     16.269   0 (0.0%)     34.595          89.688
write:      ops/s        kB/s      kB/op    retrans    avg RTT (ms)    avg exe (ms)
           10.113     326.282     32.265   0 (0.0%)     19.688       72446.246
  1. What exactly is the meaning of avg RTT (ms) and avg exe (ms)?
  2. avg exe for writes is 72 seconds(!) -- would you say this is abnormal and, if so, how do I go about troubleshooting this further?

I'm using NFS over TCP. Both the client and the server are on the same GigE LAN.

linux nfs networking

1 Answers

    1. AFAIK RTT is the server response time, e.g. from the time the client sends a read/write request until it receives an ack from the server saying "thanks, got your request, putting it on the queue". exe is the entire time from the sending of the request until IO has been done and copied to/from the client.

    2. Depends. We have some fairly loaded NFS servers with even higher "exe" times. OTOH writes are often asynchronous so large exe times don't necessarily mean the system is completely unusable.

    • 2