May 24, 2019 at 4:41 pm
While analyzing an issue, I stumbled on to a "FlushCache" error message (error log). There were many (both AM + PM). And this had been going on for quite a while. Had not seen this message before. First instinct was to check other production SQL Servers (VMs) to see if this was happening else where. That was not the case.
Next I checked the server memory, SQL memory, buffer pool and SQL Server max memory. No issues. SQL max memory was more than double the buffer pool. There was adequate memory for the OS as well. BTW - this is SQL Server 2014 SP2 / STD Ed.
Next target / suspicion area is storage. I'm running perfmon right now for both logical and physical disk information.
Any ideas for what could be causing this?
Any comments / suggestions / urls are appreciated.
Sample error log content:
2019-05-14 09:41:49.88 spid15s FlushCache: cleaned up 7506 bufs with 6810 writes in 220866 ms (avoided 5396 new dirty bufs) for db 10:0
2019-05-14 09:41:49.88 spid15s average writes per second: 30.83 writes/sec average throughput: 0.26 MB/sec, I/O saturation: 359, context switches 5417
2019-05-14 09:41:49.88 spid15s last target outstanding: 2160, avgWriteLatency 17
Thank you.
May 24, 2019 at 5:48 pm
The message is from a long checkpoint where it exceeds the recovery interval. If the memory looks fine then it's usually related to slow IO. In addition to the perf mon counters, you might want to also check sys.dm_io_virtual_file_stats.
There is a blog post about the message...it doesn't say much other than what I just said about long checkpoints but here is the link:
How It Works: When is the FlushCache message added to SQL Server Error Log?
Sue
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply