disaster recovery ensemble node

[vladimir_v]

Hi,

I have Fabric cluster with three root containers (which are ensemble members as well) and each of them has one child container running A-MQ.

By accident I started node 1 with "bin/fuse" instead of the wrapper script and it deleted my fabric configuration on that node:

17:41:03,914 | INFO | admin-1-thread-1 | FabricConfigAdminBridge | 67 - io.fabric8.fabric-configadmin - 1.0.0.redhat-379 | Deleting configuration io.fabric8.zookeeper

17:41:03,951 | INFO | admin-1-thread-1 | FabricConfigAdminBridge | 67 - io.fabric8.fabric-configadmin - 1.0.0.redhat-379 | Deleting configuration io.fabric8.zookeeper.server.a9fb7044-44d0-45b9-992a-ad94cc71334f

17:41:03,980 | INFO | admin-1-thread-1 | FabricConfigAdminBridge | 67 - io.fabric8.fabric-configadmin - 1.0.0.redhat-379 | Deleting configuration io.fabric8.agent

17:41:04,007 | INFO | admin-1-thread-1 | FabricConfigAdminBridge | 67 - io.fabric8.fabric-configadmin - 1.0.0.redhat-379 | Deleting configuration io.fabric8.version

A few hours later I successfully recovered the node with help from http://www.tuxtrix.com/2013/08/how-to-recover-fuse-esbmq-fabric.html

My question is how can I do complete disaster recovery of a failed node? If my hardware die and want to re-install one ensemble node from scratch.

I was thinking to resize ensemble to 1 node and then do fabric:join but that's not possible if one of the ensemble nodes are not on-line.

Thanks

[bibryam]

Hi Vladimir,

try this:

start a brand new fabric server and do fabric:join by using force option and also giving it a name that matches the old fabric's name.

[vladimir_v]

Hi Bilgin,

That did the trick. Also the path where fuse was installed needs to be the same.

Thanks a lot