Replies: 2 comments
-
Thank you. |
Beta Was this translation helpful? Give feedback.
0 replies
-
Hi,
One way to do this is with the |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Hello,
I noticed that every now and then we get runaway queries on our Crate cluster. For example this afternoon,
reported that several queries were running since 11am. Most of queries listed there would normally execute in less than a second, so it looks like the query jobs could be deadlocked. For the record, here's one of the queries:
We have a 3-node K8s-managed deployment with official Crate
4.1.8
Docker images and we noticed that just before those queries started running some of the Crate processes ran out of memory and died, then K8s restarted them. I combed the logs for any clue about why those queries got stuck but had no luck. I don't know if this is relevant at all, but when I killed the query jobs, somekill
statements reported that two rows had been deleted even thoughselect * sys.jobs
had shown one row for that job ID, for exampleAny idea what could cause this and how to prevent it from happening? Is there a way to automatically kill jobs that have been running for hours? Also, are those runaway queries tied to actual objects on the Java heap so the GC won't be able to reclaim memory until the corresponding job is killed?
Thanks alot!!
Beta Was this translation helpful? Give feedback.
All reactions