Curious anyone running ES on CoreOs? I have found around some unit files... But how well does CoreOs recovery/HA model work with ES recovery/HA? If you lose a node ES will go into recovery mode, replica shards will be made master maybe some shards will be moved around etc...
Now a random node reappears on the cluster won't this have adverse effect?
I suposed we would have to restrict CoreOs from moving the node to another machine. You would also want ES to recover it's local data folder when possible to avoid having ES shuffle shards around especially if they are a fairly big size.
Yeah but what a bout the data volume? It seems people sugest to run separate core os/docker container for data. Plus what if the node shows up on a host that already has its disk already full to certain capacity?
I guess we have to be able to tell core os to have to meet certain criteria before rebalancing the nodes. If it's just a reboot you don't want to move the ES container since it's just a case of upgrade/reboot or minor network glitch. And massive rebalancing happens when the machine that went offline is still good and still has the data.
Technically we don't support CoreOS and I am not sure if we have even looked into things like this on this platform to really be able to provide any guidance.
There's been some some time since this was posted, I wanted to see if you were able to get this up and running? I'm trying to install Elasticsearch on coreOS but having trouble getting up and running.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.