Does anybody know, what the "cifs_other_ops" performance counter has for a meaning? I understand, cifs_read_ops and cifs_write_ops, but cifs_other_ops doesn't tell me anything. After googling, I found some answers, which means that this other_ops are some metadata operations or something. But I didn'f found an answer satisfying me.
The source of my questions is the circumstance, that I have on a volume a very high count of cifs_other_ops (2000-2500) and just 20-100 cifs_read or cifs_write ops in the same time. Typically, the value of cifs_other_ops begins to rise when user begins with their work in the morning and decreases in the evening, when they leave the workplace.
You might do a packet capture of the activity (either with a sniffer or pktt) and use something like Wireshark to tell you what's going on. Not sure if an mtime type request comes in as a "read" or "other", but could be some similar activity as jjeziorny mentions.
I get this. There's got to be a description of what is contained in the categories of "other_ops" "cifs_other_ops" "nfs_other_ops" and san_other_ops" and even "flexcache_other_ops".
We had a complete loss of CIFS on Friday and I'm trying to work out what happened. It looks like a recurrant visitor "cifs.stats.pBlkExhaust:info]: CIFS: All CIFS control blocks for the STANDARD pool are in use." but what are the high peaks of "cifs_other_ops" and are they related?
I'm experience the same in one of our CIFS shares, this is a volume used by the Tridion CMS application, i constaly have around 1000 read_ops and over 10000 other_ops. I'm waiting a response from the application developer on why this is happening but my assumption so far is that being a CMS it spends a lot of time checking all files to see if they have changed in order to decide if it needs to update it's cache or not.
So, whilst I can't answer your question it seems we are on the same boat. I'll keep you posted on our findings. Are you using a CMS application by any chance?