none
Memory use grows and does not shrink RRS feed

  • Question

  • We have been trucking along just fine for a while.  Now it appears that our memory usage keeps growing and I cannot figure out what is hogging memory.  The usage graph keeps growing and not shrinking over several days.  We recently scaled up hoping this would help, but it did not the usage grows over several days and then we start getting the following type errors.

    Error : Could not get the memory grant of 183528 KB because it exceeds the maximum configuration limit in workload group 'SloDWGroupMedium'

    I have googled and I am not finding much to help us out.  It appears there gets something stuck in memory, but I cannot figure out what it is.  Any ideas?

    Sunday, June 30, 2019 10:25 AM

All replies

  • Hi Dorky_D,

    Are there any long running queries or does this instance support a reporting platform where multiple users are running reports, etc.? Can you please start with this document to investigate what activity is creating this issue (link).

    The folloiwng will help you with additional workload insights for advanced performance tuning with SQL Data Warehouse (link)

    Regards,

    Mike

     

    Monday, July 1, 2019 3:01 AM
    Moderator
  • There is not a reporting platform.  There are no real long queries.  There may be a load that lasts 5 minutes +/-, but nothing longer than that.  That load may read from a staging table and then place data into a production table.  We are using Informatica Cloud do do this data movement.

    I ran the memory query, but it did not show me much more than I already knew that memory is being used and held on to.

    The Transaction log query had 120 entries with a size of 0.0028 to 0.2116 GB for a total of  12 GB.

    The transaction log rollback query was not informative to me as it did not appear to hav anything meaningful in the results.

    Wednesday, July 3, 2019 7:16 PM
  • Do you have an Azure Support Plan (link). Please file a Support Request to have this issue investigated in greater detail. If you do not have a support plan, please send an email to AzCommunity and include your Azure Subscription ID where instructions will be returned to have a support requested created.

    Thank you,

    Mike

    Friday, July 5, 2019 11:50 PM
    Moderator
  • Hi Dorky_D,

    I want to follow up on this issue to see if you managed to contact support or if you are still stuck? At the very least could you send me your instance name and I could have a service engineer take a peek at the node/cluster your instance is deployed to do a general health check. Please send to AzCommunity.

    Thank you,

    Mike

    Wednesday, July 10, 2019 11:11 PM
    Moderator
  • Hello Dorky_D,

    Have you tried changing the resource class for the user?

    Also can you post a screen shot of the memory utilization or dump the dmv data here?

    Regards

    Tuesday, July 16, 2019 9:24 AM
  • I sent an email with some details.
    Thursday, July 18, 2019 12:30 PM
  • Thank you, I see it. I will update you shortly.
    Thursday, July 18, 2019 3:11 PM
    Moderator
  • I have escalated but still no response from the product group.
    Monday, July 29, 2019 7:31 PM
    Moderator
  • I have filed an internal work item to have this investigated. The best means for having this issue looked at is via Azure Support. If you do not have an azure support plan, I can have a one-time free support ticket created to have this investigated. Please send me your Azure Subscription ID to the email alias provided previously and I will return instructions to have this investigated.

    Regards,

    Mike

    Wednesday, July 31, 2019 12:20 AM
    Moderator