[
https://issues.apache.org/jira/browse/KUDU-871?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15710962#comment-15710962
]
Todd Lipcon commented on KUDU-871:
----------------------------------
Had a cluster where a bunch of tablets got into this state after running a
buggy debug build. I managed to fix things up using this awful oneliner:
{code}
kudu cluster ksck ve1116 | perl -n -e 'BEGIN { $/ = "\n\n"; } my $leader; if
(/(ve....).*LEADER/) { $leader = $1; } else { next; } if (/(ve....).*bad
state/) { $bad = $1; } if (/Tablet (.+?) /) { $tablet = $1; } print "$tablet
$leader $bad\n";' | (while read tablet good bad ; do ./build/latest/bin/kudu
remote_replica copy $tablet ${good}.halxg.cloudera.com
${bad}.halxg.cloudera.com -force_copy ; done)
{code}
> Allow tombstoned tablets to vote
> --------------------------------
>
> Key: KUDU-871
> URL: https://issues.apache.org/jira/browse/KUDU-871
> Project: Kudu
> Issue Type: Sub-task
> Components: consensus
> Affects Versions: Public beta
> Reporter: Mike Percy
> Assignee: Mike Percy
> Priority: Critical
>
> We may need to support voting for tombstoned tablets to avoid availability
> issues in certain scenarios. See the remote bootstrap design document at
> https://docs.google.com/document/d/1zSibYnwPv9cFRnWn0ORyu2uCGB9Neb-EsF0M6AiMSEE/edit#heading=h.ftfifcm5i562
> under "Should a server be allowed to vote if it DOES_NOT_EXIST or is
> DELETED?" for discussion.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
