O1OD1 excessive checkpointing
12 Nov 2018 10:32:15 UTC
Topic 216902
(moderation:
I just noticed that there is a lot of write activity on the drive on this computer
When I check the resource manager I see that all 4 O1OD1 tasks constantly write 100s of KB to checkpoint files. I have always had the "Request tasks to checkpoint at most every:" preference set to 600 (10 minutes) to reduce wear on my SSD. I use the web preferences not the local ones (but that is also set to 600).
Is anyone else seeing the same thing or is it a problem with my preferences?
Logforme wrote:I just noticed
)
I use 900 seconds, 15 minutes and don't notice the problem. I also tweaked my Windows settings after changing to SSD drives using these ideas: https://www.maketecheasier.com/12-things-you-must-do-when-running-a-solid-state-drive-in-windows-7/ The page isn't new so not all settings are in there but the ones that are seemed to help alot.
I looked at this activity
)
I looked at this activity based on your comment. My Windows Task Manager shows a consistent 0.1 MB/s disk activity for each job. I looked at the results log file and what I see are 8000 progress marks ("."). This seems to be a standard in Einstein log files for many years. But 8000 progress marks works out to 1 every 2 seconds (jobs are lasting 16,000 seconds). I've never seen that many progress markers before, maybe it is for debug.
Also in the file is the occasional ".c" which I believe marks a checkpoint. I see 7 or 8 checkpoints (every 2000 secs), even though like you I have my checkpoint set to 600 seconds.
It seems like a excessive number of progress markers, but in the big picture, my overall disk activity is less than 1%.