I've recently been given the task of setting up a new 4 node Mapr 5.1 node in order to perform a trial run of a MapR 5.1 -> 5.2 upgrade before performing said upgrade on our existing development environment MapR 5.1 cluster. Unfortunately, I've run into an issue wherein each of the four nodes do not start MapR-warden properly, despite the fact chkconfig --list |grep -i mapr clearly shows mapr-warden set to start on runlevels 3, 4, 5:
[root@ip-10-45-100-30 ~]# chkconfig --list |grep -i mapr
mapr-warden 0:off 1:off 2:on 3:on 4:on 5:on 6:off
mapr-zookeeper 0:off 1:off 2:on 3:on 4:on 5:on 6:off
I have confirmed that mapr-warden is set to startup after mapr-zookeeper in /etc/rc3.d/
Mapr-zookeeper starts up properly on the three nodes that are configured for it. However, what is curious is that, despite service mapr-warden status reporting "WARDEN not running", if I merely issue "service mapr-warden start", it starts up without incident and /opt/mapr/conf/hostname is generated, whereas before it was not present. On #3 in our 4-node cluster, if I start warden, I am then able to start the mapr-cldb service, but not before, seeing as the previously-mentioned /opt/mapr/conf/hostname file is missing.
Does anyone have any idea what could be causing this anomalous behavior? Is this a known bug in the build revision of MapR 5.1 that we are currently rolling out via our Ansible playbooks? warden.log does not have anything in it from boot-up time, so I'm thinking it's not even trying to start, despite chkconfig saying otherwise, but if anyone has a more informed idea, I am all ears.