This is an excellent step in the right direction, and I'll probably be able to do what I wanted w/ this data. However, I was hoping the CP type would include wether the CP was continuing during the sample interval, and specifically, if it was blocking.
We have an occasional scenario where our source filer performance is being affected due to semi-sync snapmirrors not being able to keep up w/ write load due to latency between the source and destination. I'd like to be able to monitor for this situation, and take programmatic action to reduce the performance impact (temporarily quiesce the snapmirorrs).
Using the data at hand, I can look for deferred back to back cps which tells me that there is write blocking, but doesn't tell me how much. 1 or 2 seconds of write blocking doesn't cause any real noticable impact, but when we see 30-40 seconds of write blocking in a given minute, that's a problem.
Is it possible to get visibility into the amount of time the filer was blocking write requests between deffered back to back cps?
