Node will not leave cluster
Austin at vimeo.com
Wed Jun 6 11:03:01 EDT 2012
I'm no expert, however, I using riak (1.1.1 2012-03-07) RedHat x86_64.
I typically issue the leave command for the node. When:
-- all it's Ownership Handoffs (riak-admin ring_status) are done
-- it is at 0.0% in both Ring and Pending in `riak-admin member_status`
-- tailing the log on the node ( /var/log/riak/console.log ) stops showing handoffs
Then I take it to be done leaving. The documentation says, "When all data is handed off, the Erlang VM process eventually exits." But in my experience with this version, that never happens. Maybe I'm just too impatient?
So I do that manually:
-- stop the node's riak on the node with `riak stop`
-- from another node, mark it as down (riak-admin down riak at xxx.x.x.xx<mailto:riak at xxx.x.x.xx>)
Then it stays in the down list of hosts.
I'm probably doing this all wrong and just clarifying how NOT to do it. :-P
On Jun 6, 2012, at 10:26 AM, David Greenstein wrote:
I've issued the riak-admin leave command on one of my nodes but it does not seem to ever leave the cluster of 4 nodes. I've opened up all firewall ports to make sure the repartitioning isn't being blocked. riak-admin status on all nodes shows the node in question it is still connected and a ring_member. If I issue the riak-admin leave command again the error states the node is already in the process of leaving. I've made sure no clients are connecting to the node that left as well. I'm also still seeing bit cast Merge log messages in the console.log of the node that left.
Has anyone experienced this issue and have some clues as to possible causes?
riak-users mailing list
riak-users at lists.basho.com<mailto:riak-users at lists.basho.com>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the riak-users