Dinesh Bhat has posted comments on this change.

Change subject: Avoid extra fsyncs of tombstoned tablets during startup
......................................................................


Patch Set 3:

(1 comment)

http://gerrit.cloudera.org:8080/#/c/4915/3/src/kudu/tablet/tablet_metadata.cc
File src/kudu/tablet/tablet_metadata.cc:

Line 173:       return Status::OK();
I wonder if we can narrow the 'skip-fsync' to tombstoned tablet bootstrap 
situation alone. i.e, could there be more situations where we may find the data 
state sync useful ?
1) Catalog_manager trying to tombstone the tablet because of a config change 
2) Tablet copy happening either on a healthy replica or tombstoned replica
3) ExternalMiniClusterFsInspector::CheckTabletDataStateOnTS called from a gtest 
right after it deleted an empty tablet (this is test anyways, so prolly not 
that significant)

Further, I don't know if it is a problematic situation or not, but it appears 
like we may have a window wherein we can have a superblock with healthy state, 
but without any Log and/or consensus-metadata. I guess as long as there is a 
path to clean them up later, it's probably okay to have that state, isn't it ?


-- 
To view, visit http://gerrit.cloudera.org:8080/4915
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I60cb184b8de2a6a381371ddcf2fb938a19757eec
Gerrit-PatchSet: 3
Gerrit-Project: kudu
Gerrit-Branch: master
Gerrit-Owner: Todd Lipcon <[email protected]>
Gerrit-Reviewer: Dinesh Bhat <[email protected]>
Gerrit-Reviewer: Jean-Daniel Cryans <[email protected]>
Gerrit-Reviewer: Kudu Jenkins
Gerrit-Reviewer: Mike Percy <[email protected]>
Gerrit-Reviewer: Tidy Bot
Gerrit-Reviewer: Todd Lipcon <[email protected]>
Gerrit-HasComments: Yes

Reply via email to