June 10, 2019 at 2:25 pm
This error has occurred 2-3 times each night for the past 4 nights
CImageHelper::Init () Failed load of symbol file with path = A:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\Binn\;A:\Program Files\Microsoft SQL Server\MSSQL13.MSSQLSERVER\MSSQL\BINN Error Code 0xc0000004
This is SQL 2016 enterprise 16 cores, 384 GB RAM. It's a three node cluster in an availability group. This is the primary server.
A web search turns up little, shut of force encryption (it is not turned on) in the network protocols, and is maxsize set on any database.
The backups/checkdb's, etc are running during this time.
The A: drive has 7 GB free out of 10 GB total. The only thing on this drives are the SQL binaries. Nothing else.
Backups go to a network share.
There is nothing in the windows error logs, but these messages appear frequently during this period:
FlushCache: cleaned up XXX bufs with XXX writes in XXX ms (avoided XX new dirty bufs) for db XX:X
Michael L John
If you assassinate a DBA, would you pull a trigger?
To properly post on a forum:
http://www.sqlservercentral.com/articles/61537/
June 10, 2019 at 9:52 pm
Is it generating any dump files, any listed in in sys.dm_server_memory_dumps?
Sue
June 11, 2019 at 1:13 pm
Is it generating any dump files, any listed in in sys.dm_server_memory_dumps? Sue
That's one of the baffling things. There are no dumps, and nothing in the windows event logs.
The flush caches are troubling, these are new, at least with this frequency. I spent some time digging into this server after I posted this, and there has been a significant increase in deadlocks, among other things. We have deployed quite a bit of new code into the databases.
Michael L John
If you assassinate a DBA, would you pull a trigger?
To properly post on a forum:
http://www.sqlservercentral.com/articles/61537/
June 11, 2019 at 4:54 pm
That is weird...the first error is usually when a dump happens. But the long checkpoints would be a concern. I'd probably check for any other signs of disk latency or resources getting taxed around the time your getting those flush log messages. And how backups are performing during that time. They could be relatively benign though too - if there were some major loads going on at the time. Or maybe the storage system is getting pegged at that time. It's not always disk i/o but I'd probably start checking into that. Keep posting back here if you find something out - I'm pretty curious. What the other things you're seeing other than the deadlocks?
Sue
June 11, 2019 at 6:10 pm
That is weird...the first error is usually when a dump happens. But the long checkpoints would be a concern. I'd probably check for any other signs of disk latency or resources getting taxed around the time your getting those flush log messages. And how backups are performing during that time. They could be relatively benign though too - if there were some major loads going on at the time. Or maybe the storage system is getting pegged at that time. It's not always disk i/o but I'd probably start checking into that. Keep posting back here if you find something out - I'm pretty curious. What the other things you're seeing other than the deadlocks? Sue
The backups have not changed for quite a while. The last change I made to the backups was to write them directly to the network share as opposed to writing them locally and copying them. We added NIC's to the servers and a new subnet dedicated solely to backups.
The FlushCache's have stopped. I "enabled" indirect checkpoints by setting the recovery interval to 60 from 0 on the databases. That's the default in 2016, but since all of these were moved from 2012, it was not set in that manner.
I plan a re-boot of this during the maintenance window this weekend. I expect that it will fix itself.
Michael L John
If you assassinate a DBA, would you pull a trigger?
To properly post on a forum:
http://www.sqlservercentral.com/articles/61537/
May 26, 2020 at 6:48 am
Hi Michael, did your reboot fix it?
May 26, 2020 at 7:16 am
This was removed by the editor as SPAM
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply