iotop; IO usage per process; nfs getattr
I am currently looking into an environment were there is a constant of 1000 nfs operations per second.
The NFS server is a netapp SAN/NAS.
When I use nfsstat I see that 60+% of the operations are "getattr".
Using the option on the netapp to see per client statistics I see which servers are responsible.
Note that read and write IOPS are together around 10-20 IOPS.
I am now trying to determine which processes is doing these getattr NFS calls.
For this I tried to use iotop. However sshd does most IO at the moment. I believe thats because the general read/write IO is low.
Does iotop only take read and write IO into account(I believe so otherwise I can't explain why there is an average of 1000 nfs ops/second)?
Will I be able to determine which process is the cause using iotop? If no, will I be able to do this with another tool? And which one?
The distro is RHEL5.6.
Thanks in advance!