Closed josere closed 2 years ago
Hi @josere, yes, generally speaking Microsoft owns memory troubleshooting for system pools. However, there are some scenarios where you may want to be aware that a system pool is out of memory. In some cases, solving the OOM will require customer action. For example, the SloHkPool system pool manages memory for In-Memory OLTP (Hekaton). If this pool runs out of memory, it likely means that the size of your memory-optimized tables is too large for the compute size, and you will need to either scale up to delete some data. Similarly, if the InMemQueryStorePool is out of memory, you may need to change Query Store capture policy to capture fewer queries, or parameterize queries to reduce Query Store memory consumption, or scale up to have more memory in this pool.
That said, if you want to focus on the user pool only in your alerting, you can certainly modify the query as proposed (and remove the resource_pool_type
column from the SELECT clause).
@dimitri-furman Thank you for the detailed response on the issue.
@josere Please let us know if you have any further query.
@josere Awaiting your response on this.
When calculating the oom_per_second metric, should I care about system pools? Shouldn't that be Microsoft's job?
We probably want to filter out non user pools:
Document Details
⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.