I was experimenting with S3 (AWS Certified Solutions Architect – Associate (SAA-C02) – Set up Cross-Region S3 Bucket Replication Hand-on Lab) Replication and ran into strange behavior:
1. I’ve uploaded file1 to bucket1 which has a replication rule to bucket2.
2. file1 got replicated to bucket2.
3. I’ve uploaded file2 to bucket2.
4. file2 didn’t get replicated to bucket1 since there’s no replication rule set up in bucket2.
5. I’ve uploaded file2(v2) to bucket1.
6. file2(v2) didn’t get replicated to bucket2.
7. I’ve uploaded file2(v3) to bucket1.
8. file2(v3) got replicated to bucket2.
So in the end, I had:
Bucket1: file1 file2(v1) file2(v3)
Bucket2: file1 file2(v2) file2(v3)
I have a few questions:
1. It seems that bucket replication makes sure that the same number of versions exists in the replication target as there are in the origin. However, it does not make sure that those versions are identical, just the overall number. Can someone confirm this? Any further explanation would be appreciated as well!
2. Given the above, it seems that using a replication(target) bucket for anything other than replication is a bad idea?
3. Does the described behaviour have something to do with bi-directional replication between two buckets?
4. Any explanation why the file version IDs are not compared between the origin and target buckets?