Scheduled reboot for Sidekiq?

[scrappydog]

Approximately every two days my Sidekig service crashes and restarts (as shown by this memory utilization graph). It recovers quickly, but it seems like having a scheduled daily restart would be a little more graceful.... thoughts?

[michaelwittig]

What's the memory config of the sidekiq container in your task definition?

[compuguy]

I suspect your issue here @scrappydog is your single Sidekiq is getting into a OOM (out of memory) condition. I had the same issue when I deployed istoleyour.pw. If you're using the stock config with Sidekiq having only 1 gig of memory, you may need to either increase the instance size, or split out the various Sidekiq queues to separate containers. More info on how this works, and how you can split things out can be found in these articles:

• https://nora.codes/post/scaling-mastodon-in-the-face-of-an-exodus/
• https://thomas-leister.de/en/scaling-up-mastodon/

You can also take a look at my fork & branch for istoleyour.pw: https://github.com/compuguy/mastodon-on-aws/blob/istoleyourpw-deploy/mastodon.yaml