Dinesh Bhat has posted comments on this change. Change subject: [consensus] KUDU-1407 replica is not evcited when TABLET_NOT_RUNNING ......................................................................
Patch Set 3: (1 comment) http://gerrit.cloudera.org:8080/#/c/5352/3//COMMIT_MSG Commit Message: PS3, Line 7: evcited > haven't looked at this code yet, but I can't tell from the commit message w Todd, yeah this fix aimed to evict the replicas returning TABLET_NOT_RUNNING. I had looked at your comments on the JIRA before attempting a fix on this. I may have misunderstood the bug little bit here. My understanding was that tablet server would return TABLET_NOT_RUNNING error once after tablet has reached a "steady" state of FAILED or some other error state. It appears like we return TABLET_NOT_RUNNING when tablet is anything other than RUNNING, so bootstrapping would fall under this too. We definitely don't want to evict when tablet is bootstrapping (until we hit a timeout ? ). As a solution, a) would it be better to introduce another response code here ? something like TABLET_BOOTSTRAPPING which is an indication for consensus to not to evict such replicas from config until some timeout, and treat TABLET_NOT_RUNNING as a fatal error ? b) Or else we could treat TABLET_NOT_RUNNING as a transient error code and after 300 secs (sufficient window to copy large tablets given the improvements in upcoming tablet copy workflows ?) we could evict the replica from config. -- To view, visit http://gerrit.cloudera.org:8080/5352 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I554ad61f25a7de78eda60fd50228e0f015c1b625 Gerrit-PatchSet: 3 Gerrit-Project: kudu Gerrit-Branch: master Gerrit-Owner: Dinesh Bhat <din...@cloudera.com> Gerrit-Reviewer: Adar Dembo <a...@cloudera.com> Gerrit-Reviewer: David Ribeiro Alves <dral...@apache.org> Gerrit-Reviewer: Dinesh Bhat <din...@cloudera.com> Gerrit-Reviewer: Kudu Jenkins Gerrit-Reviewer: Mike Percy <mpe...@apache.org> Gerrit-Reviewer: Tidy Bot Gerrit-Reviewer: Todd Lipcon <t...@apache.org> Gerrit-HasComments: Yes
