2

Imagine I have an S3 empty bucket BUCKET. I PUT two files into BUCKET with a sub-key, but have no other interaction with it:

    PUT s3://BUCKET/x/y.dat
    PUT s3://BUCKET/x/z.dat

Now, if I make a LIST call on x, will this exhibit read-after-write consistency or not?

    LIST s3://BUCKET/x/

To be clear, at this point I will never have made any previous metadata query about x or its children. I will also never mutate x in any way after the LIST call. Am I guaranteed to see both y.dat and z.dat as the result of my call?

This question touches on the consistency model of LIST, but does not explicitly pertain to my constrained example.

teabot
  • 15,358
  • 11
  • 64
  • 79
  • To my eye, the question you linked addresses this directly. "No". "Subdirectory" paths in s3 keys are fiction -- they are just characters that are part of the key. There is no parent-child hierarchy, no file tree at all; "x/" is just a prefix filter over keys in the bucket. – ben author Jun 26 '19 at 14:38

0 Answers0