cube
cube copied to clipboard
cubestore memory leak
Describe the bug I am running a self-hosted cubestore router with docker in a Nomad cluster. Since the last upgrade, after about 10 minutes, cubestore OOMs or becomes unresponsive to readyz checks and gets restarted by the Nomad scheduler. There are 3-5 cube API instances connected to my cubestore router, doing 10-50 queries per second each. There are no preaggregations or workers set up.
Memory usage before, with v0.34.33:
Memory usage now, with v0.34.36:
Hello @schneefux ,
I made some improvements across the last releases. Could you give it a try v0.35.8?
Thanks
@schneefux Did you try the latest version? I've seen reports in the community Slack (https://slack.cube.dev) that it makes a real difference with regards to memory.