Warning!!! Don't install 13.0.24 it needs to be pulled.

DonFerguson

I just upgraded to 13.0.24 from 13.0.23 and the CPU on my repository host shot up from about 25% to 100%.    

There is a new query introduced that has a "No Join Predicate" warning in the execution plan.

SELECT

[Leaf].[Id]

, [Leaf].[IdCollectionDate]

, [Leaf].[Cluster_Name]

, [Leaf].[Cluster_SqlServer_Name]

, [Leaf].[Cluster_SqlServer_Database_Name]

, [Leaf].[Cluster_SqlServer_Database_File_Name]

, [Leaf].[Cluster_SqlServer_Database_File_Type]

, [Leaf].[CollectionDate]

, [Leaf].[Cluster_SqlServer_Database_File_UsedSize]

FROM [data].[Cluster_SqlServer_Database_File_UnstableSamples_DateRange](@MinDate, @MaxDate, @MaxSamples, @EventCount) AS [Leaf]

 WHERE 

[Leaf].[Cluster_Name] = @Cluster_Name

 AND 

[Leaf].[Cluster_SqlServer_Name] = @Cluster_SqlServer_Name

DonFerguson

So I'm not sure I fully understand this, but after the CPU being cooked for about 18 hours and SQL Monitoring showing the query I provided as the culprit, the issue has calmed down.  I assume that perhaps a better execution plan might be the answer.  But things now seem better on my end. So for anyone who wants to try 13.0.24, you may not have the same experience I had.  But proceed with caution.

DonFerguson

I recycled the base monitors connected to the specific database server that is having issue and CPU went back to 100%.  So something is going on with this release and SQL Server.  I am going to roll this one back and see if it resolves the problem.  I will first run a schema compare to see if I need to rollback schema changes as well.

DonFerguson

Two base monitors in my multi monitor config are having issues with 13.0.24 in that they are running stuff that is absolutely spiking the CPU on the repository database server and after rolling these two back to 13.0.23 the issue resolved.  But what's interesting is it's only two base monitoring systems that are having this issue.  All of the others seem to be fine on 13.0.24.  So what exactly is different on the two systems not liking 13.0.24, is something I haven't been able to figure out yet.  I guess it's time to grab the logs and submit a ticket.

daz71

Yes, seeing the same type issue. Our repository Database for SQL monitor is an Azure SQL Database instance. The DTU for it shot up since the upgrade to 13.0.24.  It's maxed out the DTU for the instance and because of this we are losing connection to the instance. SQL Monitor is now unstable due to this and not usable.

I opened a ticket this morning about this, but am yet to hear anything back.  I'll check this forum going forward before upgrading the next time.

Greg_Smulko

Thank you for the reports. I'm really sorry that this update caused issues.
We pulled this version and will investigate as a matter of priority.

Greg_Smulko

Just to give you an update, we managed to pinpoint the culprit: it's related to our work to support more hosting platforms in the Estate section (Linux, Amazon RDS, Azure MI).
We're aiming to release the fix early next week.
Apologies again!

DonFerguson

I'm glad to hear that.  Thanks for the update Greg!

daz71

Thanks for the update Greg

DonFerguson

I noticed that 13.0.25 was released.  I then updated the two systems that I had rolled back to 13.0.23 to 13.0.25 and noticed no ill effect on SQL Server performance.   After confirming that to be the case I updated the rest of the SQL Monitor hosts to 13.0.25 and all looks good.  Thanks for resolving this issue.

Greg_Smulko

Thank you for the confirmation Don!

SEarle86

Hi,
We are on 13.0.25.22151 and are seeing the same issue periodically:


sp_whoisactive:

Tom_C

Hi @SEarle86,

Appreciate we've been dealing with this on an internal ticket but just wanted to add an additional response here for anyone else who might be seeing an issue with high CPU after updating past 13.0.24

Our development team are currently investigating what they believe might be behind the CPU issue observed, although this may look similar to the query originally reported by Don on this forum post it is a different issue. If you find this forum post and are experiencing the same symptoms (on 13.0.25 or later) we would encourage you to log a ticket with support for us to gather some additional information for diagnosis.

Hope to have an additional update here soon with further information on this.

Tom_C

A fix was merged that went out with 13.0.33 (which was pulled and replaced by 13.0.34) which @SEarle86 has advised has resolved the issue seen.

13.0.34 can be downloaded here, please let us know if any issues updating or if you believe the issue persists post update please let us know.