AnsweredAssumed Answered

Reconfiguring MapR after cluster halt in EC2

Question asked by cybertextron on Apr 16, 2015
I'm currently installing MapR 4.1.1 in a EC2 cluster. I have all the scripts to configure the MapR cluster, and when I have all the services running:

    ubuntu@ip-10-10-126-46:~$ sudo jps
    
    21179 CLDB
    
    27781 Gateway
    
    28857 HMaster
    
    20435 WardenMain
    
    26142 CommandServer
    
    30995 JobHistoryServer
    
    26606 ResourceManager
    
    8371 Jps
    
    9434 QuorumPeerMain

I'm using Vagrant to manage the EC2 instances. So I'm testing the halt / resume feature. First, when I bring the cluster, It works as expected, everything correct. However when I stop the cluster and try to bring the MapR services again, it fails.
I know the IP addresses change when an EC2 instance is stopped, so I'm trying to run:

    sudo /opt/mapr/server/configure.sh -C namenode.dg.local -Z namenode.dg.local -RM namenode.dg.local -HS namenode.dg.local -N mycluster -D /dev/xvdb

I another option I tried was to execute with just the `-R` option:

    sudo /opt/mapr/server/configure.sh -R

However, Everything failed to restart the `ResourceManager`, `JobHistoryServer`, `HMaster` and `Gateway`:

    11833 FsShell
    16030 Jps
    13116 CentralConfigCopyHelper
    10510 QuorumPeerMain
    10810 WardenMain

What else I can do to bring my cluster to the state it was before? The only thing that changed was the IP addresses in each node: (`/etc/hosts`)

    10.185.145.5         datanode-1 datanode-1.dg.local
    10.185.139.72        datanode-2 datanode-2.dg.local
    10.10.126.46         namenode namenode.dg.local

and now it is:

    10.10.49.35 datanode-1 datanode-1.dg.local
    10.40.101.126 datanode-2 datanode-2.dg.local
    10.60.114.199 namenode namenode.dg.local

Outcomes