-
Notifications
You must be signed in to change notification settings - Fork 151
Closed
Description
I've noticed that the S3 uploader threads aren't able to keep up with the processing of uploading temporary files. This seemed to have happened when we change the prefix from %{+yyyy-MM-dd}/ to %{+yyyy-MM-dd}/%{type}/. From one of our production Logstash instances:
> find /tmp/logstash/ -type f | wc -l
539
I'm still investigating this, but I suspect that we need more upload workers? The specific problem here is that the logs don't give me any indication that something is going wrong.
Metadata
Metadata
Assignees
Labels
No labels