-
Notifications
You must be signed in to change notification settings - Fork 67
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Vector query consumes >115 GB of memory #5475
Comments
It looks like conditions have changed for this one. The merge of #5484 happened to help a different query (mgbench bench2/q5, querying against Parquet) improve from "OOM killed" to "finishes ok" so I figured I'd give this one a retry as well. Per the repro below, starting with commit 87ab6b7 that's associated with the changes in #5484, instead of consuming >115 GB of memory+swap and getting OOM killed, I observed it now ran for about 3 minutes, consumed a bit above 20 GB of memory, then died with the panic shown.
|
Verified in super commit 7c930f2. The query now runs successfully without panic.
Thanks @mattnibs! |
Repro is with super commit fdc2852.
Test data is
s3://brim-sampledata/mgbench/bench2.csup
.On my Macbook this query ran for 17+ minutes and Activity Monitor showed the
super
process consuming over 115 GB of memory (so, lots of swap) before it got shot by the OOM killer (I think my system ran out of free disk to hold swap by that point). At first I assumed this meant its aggregations might require spill-to-disk functionality that has not yet been added for vectors, but maybe there's something else to it, hence the issue.The text was updated successfully, but these errors were encountered: