jdef
jdef
once #417 lands, this PR should mark up API additions w/ "alpha" annotations
Is there actually a spec issue here, or is this some k8s/CSI implementation issue?
In this case it sounds like "API address" is related to the SP's storage topology, and there's a desire to map different classes (ala StorageClass) of storage to different backends/racks...
I've given this a little more thought and, you're right, the topology suggestion I made doesn't quite fit. That said, I'm wondering if you *really* want a single instance of...
> CSI hasn't actually tried to resolve tenancy issues - we've typically punted to COs for that. I'm not convinced that we should try to tackle multi-tenancy at this level....
Volume id uniquely identifies a CSI volume. I don't see anything wrong with embedding the tenant id inside the volume id if that's what your implementation needs to uniquely identify...
requires upgrade to Mesos v1 API https://github.com/apache/mesos/commit/eec19286f93e9c5b9f28447421a5721eb1605ac6
suspect that VIP via flag is fine; default for the flag would mean "there is no VIP" i don't think the onus is on etcd-mesos to define behavior of duplicate...
thanks for the PR, left some comments
this ticket could use a more elaborate story. maybe describing the expected result of pushing a PR to master or release branch (snapshot images) vs. building from a release tag...