You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Note that according to an older mail from GT (subject: "worker_slow configuration" / dated 3 December 2021), the concurrency for the slow worker that handled the batch_slow queue with the async_register and async_generate_results was 2.
We should also make sure that each of the workers gets sufficient max memory. The total available memory on each of the Docker based batch servers (dev-docker.batch and docker.batch) is currently 16 GB.
PR #1426 moves batch_slow to slow worker and reduces concurrency on that worker.
Memory for both the normal and slow worker can be configured already in local.env by overriding the WORKER_MEMORY_LIMIT and WORKER_SLOW_MEMORY_LIMIT respectively.
To solve memory issues on batch installations:
12The text was updated successfully, but these errors were encountered: