Closed rnd closed 2 years ago
I'm in favor of #2, we are running several service instances, running cleanup routine on each of them independently would probably create issues. 48 hours expiration time should be enough.
I agree, it's more reasonable to take the 2nd approach then. Not sure how we proceed with this as i don't have access to the cluster. Should i wrote the bash script to clean up the files and you put it on k8s cron manifest after?
Do we think this is blocking shipping, or can we go live on UI side and see how things go? I assume we can always cleanup manually in the interim if needed.
I don't think its a blocker. And you're right -- in case we run out of storage space because of these files, we can remove it manually until we have the cron deployed.
Found a bug or two on the UI side, so we'll get it shipped tomorrow!
closing this as Andrey already deployed it.
With the addition of resumable upload on publishing endpoint, currently it's possible for the server to run out of storage space because we have too many unfinished file upload as TUS does not handle file upload expiration. see: https://github.com/tus/tusd/issues/435
This ticket is created to discuss the possibility to solve this issue. There are 2 ways that I think can be implemented:
As for the file expiration time, I might need input from @tzarebczan or @andybeletsky as this might affect user experience and require changes / docs on the frontend too