1 Answer
- Newest
- Most votes
- Most comments
4
I have performed some research into your situation and have performed some testing in my personal lab environment. In this case, due to the current limitations of cross region replication, there are 2 options that I could think of. I have listed them below along with my thoughts:
-
- The workaround detailed in the GitHub request you mentioned. This involved creating a temp tag, deleting original image, using put-image thus triggering replication, then cleaning up the temp tags from source and destination registries in affected regions. This option is valid and there is no issue with deleting an image tag and then pushing the same tag again. Obviously, there are data transfer charges involved here as you already mentioned which cannot be avoided. Also there will be some downtime as you will be deleting images from the repos and recreating them.*
-
- The other option you mentioned is to create the destination registry, then pull down the images from the source registry and, then push into the destination registry. Then enable the additional region replication on the source registry so, moving forward, new images/tags will be replicated automatically. Only one set of data transfer charges would be needed here to push images to the new registry. Also there would be no downtime on the source registry as no images are being deleted.
I'd recommend the second option since this has less impact on users of the registry and there is less risk of things being deleted accidentally. However, since there is manual manipulation of the images there is still risk something is missed when pushing the images to the new registry in the new region. Creating a script to automate this will help reduce risk.
Relevant content
- asked 5 months ago
- asked 2 years ago
- asked a year ago
- AWS OFFICIALUpdated 4 months ago
- AWS OFFICIALUpdated 2 years ago
- AWS OFFICIALUpdated a year ago
- AWS OFFICIALUpdated 2 years ago