Alright, the results are out. :)
TLDR: I think that there is a major improvement. I still see spikes, but they are usually (not always) every 5 mins instead of the previous 1 min interval. This much may be fine (I will try to investigate further).
I tried playing around with the settings etc. What I found were:
- With a fresh install, re-creating the Pools and Datasets etc. instead of importing them, the spikes still happen, but roughly every 5 mins. Although I am not sure, I would guess that this should be when the system dataset gets written into. No writes in-between.
- After adding an app, setting up shares etc., things aren't looking as good (the app was stopped after installation to mitigate any I/O etc.). This is because at this point, the CPU spikes started happening more often. Still not as often as previously with BETA-1, but still plenty of spikes. (The red interval is when I started an I/O intensive app, which was then stopped.)
- It was mentioned in the ticket that the logging has somewhat changed. I think I can verify this, as it seems that every value is being logged a lot more frequently: the values of the dashboard are being updated much, much faster than they were on Bluefin (at least for me), and also (as a kind of drawback I might say) the disk I/O looks weird on just about any disk if there is any I/O as the logs are just too frequent (see the image below).
- Due to these "too" frequent logged values, it may be that at this point, this much is fine. I will try to measure how much power the system is using with a meter I have at home. If these bumps are in any way important, then I should be able to see them on the readings. If there are no jumps in the readings however, that would mean that this is fine now and this is now only because of the updated, too granular logging frequency.
TLDR: I think that there is a major improvement. I still see spikes, but they are usually (not always) every 5 mins instead of the previous 1 min interval. This much may be fine (I will try to investigate further).
I tried playing around with the settings etc. What I found were:
- With a fresh install, re-creating the Pools and Datasets etc. instead of importing them, the spikes still happen, but roughly every 5 mins. Although I am not sure, I would guess that this should be when the system dataset gets written into. No writes in-between.
- After adding an app, setting up shares etc., things aren't looking as good (the app was stopped after installation to mitigate any I/O etc.). This is because at this point, the CPU spikes started happening more often. Still not as often as previously with BETA-1, but still plenty of spikes. (The red interval is when I started an I/O intensive app, which was then stopped.)
- It was mentioned in the ticket that the logging has somewhat changed. I think I can verify this, as it seems that every value is being logged a lot more frequently: the values of the dashboard are being updated much, much faster than they were on Bluefin (at least for me), and also (as a kind of drawback I might say) the disk I/O looks weird on just about any disk if there is any I/O as the logs are just too frequent (see the image below).
- Due to these "too" frequent logged values, it may be that at this point, this much is fine. I will try to measure how much power the system is using with a meter I have at home. If these bumps are in any way important, then I should be able to see them on the readings. If there are no jumps in the readings however, that would mean that this is fine now and this is now only because of the updated, too granular logging frequency.
