Upgrade from DP4 -> 1495rel
CentOS 5.8 64-bit I've tried upgrading one node in a 4 node cluster from DP4 to the 1495 build, with no luck. Tried removing the node from the cluster, uninstalling DP4, installing 1495, and telling the node to join the existing cluster. Rebalance fails continually:
Service compaction_daemon exited on node 'ns_1@xxx.xxx.xxx.144' in 0.02s supervisor_cushion001 ns_1@xxx.xxx.xxx.144 08:26:15 - Wed Aug 1, 2012
Node 'ns_1@xxx.xxx.xxx.142' saw that node 'ns_1@xxx.xxx.xxx.144' went down. ns_node_disco005 ns_1@xxx.xxx.xxx.142 08:25:56 - Wed Aug 1, 2012
Haven't heard from a higher priority node or a master, so I'm taking over. mb_master000 ns_1@xxx.xxx.xxx.144 08:25:34 - Wed Aug 1, 2012
Node 'ns_1@xxx.xxx.xxx.143' saw that node 'ns_1@xxx.xxx.xxx.144' went down. ns_node_disco005 ns_1@xxx.xxx.xxx.143 08:25:30 - Wed Aug 1, 2012
Shutting down bucket "sms_fs" on 'ns_1@xxx.xxx.xxx.144' for server shutdown ns_memcached002 ns_1@xxx.xxx.xxx.144 08:25:25 - Wed Aug 1, 2012
Bucket "sms_fs" loaded on node 'ns_1@xxx.xxx.xxx.144' in 0 seconds. ns_memcached001 ns_1@xxx.xxx.xxx.144 08:25:25 - Wed Aug 1, 2012
Service compaction_daemon exited on node 'ns_1@xxx.xxx.xxx.144' in 0.00s supervisor_cushion001 ns_1@xxx.xxx.xxx.144 08:25:24 - Wed Aug 1, 2012
Couchbase Server has started on web port 8091 on node 'ns_1@xxx.xxx.xxx.144'. menelaus_sup001 ns_1@xxx.xxx.xxx.144 08:25:24 - Wed Aug 1, 2012
Somebody thinks we're master. Not forcing mastership takover over ourselves mb_master000 ns_1@xxx.xxx.xxx.144 08:25:24 - Wed Aug 1, 2012
Haven't heard from a higher priority node or a master, so I'm taking over.
Any ideas? This is a test cluster I'm using, but will need to eventually do this on a 9 node cluster. Thanks!
If the DP4 nodes and the 1495 nodes can't live together, then is it safe to assume the data and views on the DP4 nodes will be there after the upgrade to 1495?
The main reason I was trying to migrate to 1495 was to have support for cbbackup.
Since cbbackup doesn't work in DP4, I've developed a RESTful method for backing up and restoring the data and views, but if there are vast differences in the two releases, I'm weary of finding additional differences with how 1495 might perform with my application. So far DP4 has been rock solid, save for the lack of cbbackup working.
I understand it's still in development, but when a final release is made available, an upgrade path sure would be useful.
Thanks.
Yes it is safe assume the data and views on the DP4 nodes will be there after the upgrade to 1495. Right now what you can do is:
1.Download and install the latest 2.0.0 couchbase on another independent machine outside your DP4 cluster.
2.Under /opt/couchbase/bin you will find our cbbackup and cbrestore script. You can use them to do backup all the data and views from the whole DP4 cluster.
3.All the data will be stored in your backup folder.
4.After your upgrade your DP4 cluster to 1495, use the cbrestore script to do the restore job.
Since backup and restore is a client application, it can run anywhere outside your cluster. Take your data backup from DP4 and restore it to 2.0 cluster won't lose anything you have in DP4 cluster. Backup file features are not affected by versions.
So you want the 1495 node to join the existing DP4 cluster? That won' work. 2.0.0-1495 node and DP4 nodes can't exist in the same cluster. Rebalance failure will definitely happen. Maybe you should upgrade all the nodes and then cluster them :-)