[
https://issues.apache.org/jira/browse/KUDU-3311?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17448263#comment-17448263
]
ASF subversion and git services commented on KUDU-3311:
-------------------------------------------------------
Commit e3bcd743d6c2d8791591c69260e7d9e6bc59fef9 in kudu's branch
refs/heads/master from Zoltan Chovan
[ https://gitbox.apache.org/repos/asf?p=kudu.git;h=e3bcd74 ]
[master] KUDU-3311 Allow to start with diff num of masters
Modified SysCatalogTable:Load to allow cluster startup if +1
master address is present in the flags.
The startup is halted, if the number of masters on-disk is higher than
what was provided in the flags, discouraging decommissioning masters
this way.
Additional manual test was done with the following steps in order to
ensure, that the documented way of removing a master works:
- ./start_kudu.sh -m 4 -t 3 -b $KUDU_BUILD_DIR -c /tmp/data --rpc-master 8800
- looked up the "last" master which was listening on :8006 and removed
it by running:
./kudu master remove
127.0.0.1:8800,127.0.0.1:8802,127.0.0.1:8804,127.0.0.1:8806 127.0.0.1:8806
- stopped the cluster by running ./stop_kudu.sh
- restarted with 3 masters by running
./start_kudu.sh -m 3 -t 3 -b $KUDU_BUILD_DIR -c /tmp/data --rpc-master 8800
Change-Id: I39aeee2f52a55a8c29770f748895d38c9adff8a2
Reviewed-on: http://gerrit.cloudera.org:8080/17995
Reviewed-by: Andrew Wong <[email protected]>
Tested-by: Kudu Jenkins
> Allow masters to start up with a list of masters with a diff of one from
> what's on disk
> ---------------------------------------------------------------------------------------
>
> Key: KUDU-3311
> URL: https://issues.apache.org/jira/browse/KUDU-3311
> Project: Kudu
> Issue Type: Improvement
> Components: master
> Reporter: Andrew Wong
> Assignee: Zoltan Chovan
> Priority: Major
> Labels: newbie
>
> Now that Kudu automatically adds a master if we start up a new master
> alongside an existing only set of masters, we should also loosen the
> restriction that the gflag is the same as the existing Raft config, in case
> users want to add a master and then restart the entire cluster at the same
> time.
> This seems like it would be common enough for orchestration tools like CM,
> which marks the cluster as stale and suggests a full service restart upon
> adding a new master role.
--
This message was sent by Atlassian Jira
(v8.20.1#820001)
