allegroai / clearml-server

ClearML - Auto-Magical CI/CD to streamline your AI workload. Experiment Management, Data Management, Pipeline, Orchestration, Scheduling & Serving in one MLOps/LLMOps solution
https://clear.ml/docs
Other
386 stars 134 forks source link

Fix elasticsearch memory requirements #254

Closed r0mk1 closed 1 month ago

r0mk1 commented 1 month ago

Elasticsearch service silently terminates because of OOM. Add memory requirements according to the official page "Install Elasticsearch with Docker": docker run -m 1GB ...

evg-allegro commented 1 month ago

@r0mk1 According to the Elasticsearch documentation limiting memory through docker settings is equivalent to limiting memory through ES own jvm settings. And by default they suggest not to set those. In this case ES will use up to 50% of available memory on the container host. https://www.elastic.co/guide/en/elasticsearch/reference/current/docker.html#docker-set-heap-size Since you get OOM error I assume limiting the memory even more would not help. In such case we recommend first to make sure that your host has enough memory for running Elasticsearch. How much RAM does you host currently have? The memory that ES requires depends on your actual load. So increasing the host memory would likely solve the issue

r0mk1 commented 1 month ago

@evg-allegro, you're right, I removed the memory limits config and the server starts fine anyways. Apparently it was something else causing the error I saw. Sorry for the false alarm. I'm closing the PR then.