Retention policy and clustering

[Jean-Daniel]

Hello,

In my configuration, I have a single public zot instance used for pushing images, and multiple private zot instances in read-only mode used to pull the images. How should the retention policy be configured ?

I tried to configure it, but I can't see where zot stores the information about pulled/pushed images used to determine which image to retain. Is it somewhere in the S3 storage, or locally ? How is this counter shared across instances when using clustering ?

And finally, should retention be configured on all machines or only on the instances where pushing images is allowed?

Thanks.

[rchincha]

@Jean-Daniel zot itself is not clustering-aware (at least not yet) but you can set up a clustering solution which you seem to have done. Hence, retention is also node-local. Here is an outline to make this scheme work.

1. Single writer which lands on a zot instance and you can setup retention accordingly (keep tags, etc)

2. Other zot instances can be readers with sync/mirroring turned for specific repos per instance and retention can be configured to follow tags/images in those specific repos.

[Jean-Daniel]

Thank you for this reply. This solution will require to duplicate the storage, but it is a price I'm ready to pay to get retention working (and I don't need large storage with retention enabled).