Github user HeartSaVioR commented on the issue:
https://github.com/apache/storm/pull/2005
@arunmahadevan @satishd
I have updated the problem definition and fix for the problem.
If deleting blob is requested from superuser (say, Nimbus, Admin.
Supervisor is not since Supervisor is only superuser for reading) we don't need
to get meta since the user already get WRITE privileges for any blobs. Syncing
up blobs is done via Nimbus privilege so it will pass downloading and/or
reading meta.
If deleting blob is requested from user, the logic will be run via leader
Nimbus. Given that the scenario, both `ZK node not found` and `meta file not
found` are all errors which users (maybe admin) need to know about the thing.
Ideally we might want to consider repairing, but it makes the logic more
complicated so I'd rather define Blobstore H/A as not fully reliable and
recommend HDFS BlobStore for H/A environment.
Please take a look again. Thanks in advance!
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
