Replies: 1 comment 3 replies
-
Hey @lrojas94 👋 Thank you very much for this report. I want to ask a bunch of questions beforehand:
Thank you very much for helping me debug this. |
Beta Was this translation helpful? Give feedback.
3 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
First off - Hello again! I asked a question before and was really happy with you guys' support <3
So, here's another one for you.... We're on Meilisearch 1.6.
Basically, the problem is our instance keeps becoming unavailable quite regularly, specially after one of these crazy CPU spikes (We don't have a CPU limit for it). The pod it's in is a Digital Ocean 8 vCPU/32GB (7,500 IOPS/ 300 MB/s Throughput). The white-spots on the table after the spikes are times it's either reboot itself, gone down or is completely unavailable.
From what we can tell, it seems related to really high search usage during our peak hours, although we don't have a particular metric that can tell this sadly. We have a big data set (about 70GB of data) divided among multiple indexes, but our searchable attributes (I believe) are setup so it's fairly performant - quite minimal.
We have tried disabling our indexing jobs (Which did nothing at all).
Another weird thing is that - when indexing - Meili won't use more than 5GB RAM. Stop indexing and it jumps to 24gb (our max atm). Seems weird, thought it should be noted, although I can't tell how these relate since the CPU spikes / issues happen regardless of whether or not the indexing queue is empty.
On the above point - here's how the RAM chart looks like (No new queued jobs since the 20:00 mark or so)
Any help would be appreciated!
Beta Was this translation helpful? Give feedback.
All reactions