Recordings remain `started` forever

Issues with deployment
1

I have a PRO edition 2.28 deployed on AWS EC2, with recordings being stored in S3. Since in our application the sessions happen in a pre-planned manner according to a schedule, and sometimes there can be no sessions for a few days in a row – to cut costs, I have a cronjob that stops instances if there’s no sessions planned in the next 15-20 minutes.

That cron job is set up to not stop the cluster if there are any recordings still being processed, by checking if any of them has status started. Recently, our AWS bill grew up, and I found out that cronjob fails because some recordings fail to clean up and remain “started” forever.

When I check contents of /opt/openvidu/recordings, each failed recording folder just has a file like .recording.65c204727cd5c

Is it a known issue? If yes, was it fixed in 2.29?

Thanks.

2

No, it is not a known issue.

How many recordings do you have per day and how large are your instances?

Are those recordings failed attemps?

3

Max. 1-2 recordings per day, not even every day. Instances are c5.xlarge.

Recordings with those name never appear among finished recordings, so I would say yes, those are failed recordings.