Currently we're configuring ES with Ansible, passing initial cluster members with Ansible.
I wonder if GCE discovery would be of any help / improvement.
I currently would think the documentation is outdated and might be no longer valid.
Well, I don't like the concept of cluster.initial_master_nodes that much, especially because I have to apply a static configuration to a cluster, that might be outdated later on.
I think this setting cluster.initial_master_nodes should even get removed after successful bootstrapping, like only stated on [this page](Important Elasticsearch configuration | Elasticsearch Guide [master] | Elastic):
After the cluster forms successfully for the first time, remove the cluster.initial_master_nodes setting from each nodes' configuration. Do not use this setting when restarting a cluster or adding a new node to an existing cluster.
Unfortunately this role doesn't support plugins yet, so I have to either to extend our custom wrapper around this ansible-elasticsearch or extend the upstream-role to install the plugins.
Correct, these are the docs you're thinking of. But to repeat myself, this is nothing to do with discovery. For instance, you are required to set cluster.initial_master_nodes to the same value on every node, but there is no auto-discovery mechanism that can satisfy that requirement.
No, sorry, I don't have a lot of knowledge about GCE specifics and I can't see anything obvious in the code. It gets info about the network interfaces if you use the _gce:..._ special values for network.host etc, but shouldn't be updating anything AFAIK
Can you forbid that specific permission? If so, does it break? Can you share logs and/or stack traces so we can see where it's being used?
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.