1 minute read

Say you have a bucket called junk with "directory tree" (not really but…) several levels deep (junk/lev1/lev2) and want to do some ILM stuff on the stuff in one of those subdirectories.

Bucket junk with one or more levels of subdirectories

Of course, you'd create a new ILM rule and apply it in a policy.

To narrow down potential matches, administrator creates a policy for this tenant (named s3) and the bucket named junk. In it, we say if object name (aka key) contains lev1/lev2/, match it.

Bucket junk with one or more levels of subdirectories

Put together, that becomes junk/lev1/lev2/. Now, that's a CONTAINS condition, so you'd better hope you didn't forget that last / and at the same time have something like junk/lev1/lev2-IMPORTANT/ in there.

Anyway, once that's in place we need to decide want to do with the stuff that matches this and create a policy to include this rule.

Before we put that policy in effect we can simulate it, which shows that junk/lev1/lev2/junk.file matches the rule.

Bucket junk with one or more levels of subdirectories

You may want to ensure no other stuff gets caught in this rule and that the same tenant, s3, knows what kind of matching is in place. The rule only matches against that tenant and that bucket, so other users and other buckets would not be affected.