Bug #4619
closed
- Status changed from New to Fix Under Review
Code looks good, assuming the tests run.
Sorry about that! :(
I think this isn't correct. If we restart the table server MDS, it will send two ready messages to the table client. One by MDS::handle_mds_recovery(), one by MDS::recovery_done(). I think it's better to call MDS::recovery_done() when bringing up a fresh cluster.
- Project changed from Ceph to CephFS
- Category deleted (
1)
- Status changed from Fix Under Review to Resolved
commit:968c6c0c9408b33904041e5ddbd9ea738e831713
- Status changed from Resolved to In Progress
Sage said he'd look at the double-send as well.
- Priority changed from Urgent to High
- Status changed from In Progress to Fix Under Review
recovery_done() breaks on a fresh machine because of the populate_mydir() ordering. The problem is that both recovery_done() and handle_mds_recovery(who) will catch this case, since the recovery_done() sends to everyone who is active.
I think a simpler fix is to handle the create/start case separately in boot_create()... where there is also handling for hte talbeservers. See wip-4619...
Code looks good, Sorry for the bug!.
- Status changed from Fix Under Review to Resolved
Merged and pushed to master in commit:3842ff7d677bae98462f7d050f5fda9d85f6273d
Also available in: Atom
PDF