On 28 Aug 17:37, Rhodes, Richard L. wrote:
I’m looking at the Performance Manager graph for a volume that is a CIFS share.
It graph has 4 stats:
cifs_read_ops,
cifs_write_ops,
cifs_other_ops,
cifs_ops.
CIfs_reads_ops and Cifs_write_ops are VERY few totaling only 0-30/sec, but mostly at 0.
Cifs_other_ops is a constant 130/s to 180/s, but NEVER dropping below a floor of 130/s.
Cifs_ops is similar to Cifs_other_ops, but somewhat higher.
The help metrics don’t exactly help explain these.
If you look in the API docs, there is some documentation, I believe - at least, thats where I found it.
Q) What is cifs_other_ops and cifs_ops?
I had though that cifs_ops was the cumulative value of all the ops you mentioned, but your numbers dont add up, so perhaps not.
cifs_other_ops, my understanding is that they are stat information aka metadata aka file information.
Q) What would a CIFS client be doing to cause few reads/writes,
but lots of cifs_other_ops and cifs_ops?
My guess would be something is heavily accessing the file tree. In the unix world this would be someone running 'ls -l --color' or find a lot - something that accesses the file information (creation date, mondification, permissions) for lots of files.
Hope that helps!