August 7, 2019 at 4:31 am
I'm running SQL 2014 SP3 on Windows Server 2012 R2 Std Edn and after July 2019 patching last week I'm seeing lots of errors relating to memory issues in one of two SQL 2014 instances i.e. There is insufficient system memory in resource pool 'internal' to run this query
This is causing alot of problems executing basic queries and backups are failing
The server itself has plenty of available memory and max memory on the affected instance is set to 65 GB. I noticed from a monitoring tool that we use a severe degradation in the maximum workspace memory just after the patching and an increase in pages allocated to memoryclerk_SOSNODE.
This all occurred after installed July 2019 updates KB4504418, KB4507448, KB506996 & KB507005
Do you think the updates have caused a memory leak, We are planning to install SP3 CU3 later this week
August 7, 2019 at 3:22 pm
I did see this (https://social.msdn.microsoft.com/Forums/sqlserver/en-US/60d20f7a-8d64-48b1-ade9-8ba648e5945c/there-is-insufficient-system-memory-in-resource-pool-default-to-run-this-query). I might reduce my memory slightly and give the OS more and see if that helps./
August 7, 2019 at 11:53 pm
Thanks Steve, the server a high amount of available memory so that doesnt seem to be the issue, I'm wondering what might cause the MemoryClery_SOSNode to increase massively
August 8, 2019 at 12:06 am
There's clearly a memory shortage somewhere. There are lots of memory clerks in the SQL OS, so I was thinking to reduce this to be sure the max memory is available. I'm sure the system rebooted after patching, correct?
I did see this. Does it apply? https://support.microsoft.com/en-us/help/3174708/a-memory-leak-occurs-when-data-consistency-check-is-being-executed-for
Also, do you have SQLCLR or CLR-ish (spatial, etc.) things running? Any appdomain unloads in the errorlog?
August 8, 2019 at 2:06 am
Hi Steve, thanks again. No CLR enabled and system was rebooted after patching, we also applied CU3 last night and rebooted and issue still occurrs. No consistency checks were running either. Seems like a memory leak
August 8, 2019 at 3:39 pm
I have seen a few more reports of strange bugs lately with MS than I'd expected. In this case, I might look at a CSS case with them. I'm not sure if you'll get far, as WS2012 R2 is out of mainstream support and this feels more like a mainstream rather than extended support issue. I think ES ones are mostly security.
You could try wit them. The internal pool, as I understand it, is system processes that fall under the (usually disabled) Resource Governor. It's always running on systems, though usually with just internal and default groups.
The other thing you might try is posting a note on Twitter under #sqlhelp with something like: #sqlserver 2014 increase in pages allocated to memoryclerk_SOSNODE after July patches. #sqlhelp
You could link back here and maybe get someone's attention that is more into the internals. That, or send a polite note to Paul Randal at SQLSkills. This might be the type of issue he'd write about in his newsletter and give you a quick answer.
August 8, 2019 at 11:31 pm
Thanks a million Steve for your responses.
I think the next step is a case with MS, according to the history of this server similar symptoms appeared a few months ago and was resolved by installing the latest SP at the time SP3 as there was a specific fix related to a memory leak in the MEMOBJ_SOSNODE which was release in SP1 CU4. Unfortunately, it looks like it is happening again
August 9, 2019 at 2:37 pm
Let us know what happens. I sent a note to MS, but no response as of yet.
Viewing 8 posts - 1 through 7 (of 7 total)
You must be logged in to reply to this topic. Login to reply