…I did not hardcode the dtrace calls inside NFS – Block Size Monitor. Talking to a co-worker about create a utility to see the latency for each NFS call, i did tell him that it is not needed, because he could just write another “D script” and pipe to NFS – Block Size Monitor to see the results. Obviously, i did ask him to post it here in case you find it useful too. ;-)
If you execute it like that:
/home/leal/nfsio-lat_LITE -u | /home/leal/blockmonitor
You can see the results like 1kb being 1ms! And to see if there is really strange IO on the storage, use “-m“. So, you can see if there is a “crazy” IO taking more than 1 second!