balancing numa nodes

  • Can anybody advise on how i go about balancing threads to each node. As you can see from the below screen shot node_id 2 & 3 and 6 & 7 are offline (another query i ran shows them as OFFLINE VISIBLE). This is sql2019 STANDARD so limited to 48 online schedulers which are currently spread across 4 nodes (12,12,12,12 = 48). I think it would be better to have all 8 nodes online with (6,6,6,6,6,6,6,6 =48) but not sure of the numbers to put in the ALTER SERVER CONFIGURATION SET PROCESS AFFINITY CPU in order to get everything 'balanced'

    scheduler nodes

  • Thanks for posting your issue and hopefully someone will answer soon.

    This is an automated bump to increase visibility of your question.

  • i don't know if my post needs for info for an answer but i tried running the following. I would have thought the section where it says CPU = 0 TO 5 would have given me 6 cpu's with 6 schedulers, but it appears to give me 10 and 10 respectivley ? If somebody can explain the reason behind this I might be able to logically work out what I need to put in - its totally unbalanced using the numbers in my query (but at least I got all 8 nodes to show ONLINE)

    QUERY to alter cpu schedulers

  • Figured this out by running:

    -- SQL Server NUMA node and cpu_id information
    SELECT parent_node_id, scheduler_id, cpu_id, status
    FROM sys.dm_os_schedulers WITH (NOLOCK)
    WHERE [status] IN (N'VISIBLE ONLINE', N'VISIBLE OFFLINE')
    ORDER BY parent_node_id, cpu_id;

    and then working out the id's of the cpus that needed bringing online for the corresponding parent_node_id's

Viewing 4 posts - 1 through 4 (of 4 total)

You must be logged in to reply to this topic. Login to reply