I have a scenario where we created an Elasticsearch cluster using a base Ubuntu 16.4.0.4 VM template. These Elasticsearch nodes are able to be SSHed into using their host names (ES-DATA-01.domain.com). I didn't think anything of it until I created another VM utilizing the same 16.4.0.4. VM template without Elasticsearch, and I am unable to SSH using the host name.
I was curious about this behavior and investigated further. /etc/hosts/ is configured appropriately on both machines to resolve to their FQDN, and /etc/dhcp/dhclient.conf is also identical across both of these VMs. ifconfig doesn't show any differences between them either.
What Linux configuration(s) does a base Elasticsearch install alter in order for SSH to work by using the host name? I am going out on a limb that within elasticsearch.yml, the config value of node.name: ${HOSTNAME} is playing into this by some sort of pass-through mechanism, but I cannot ascertain the mechanism. Perhaps unicast is also a factor here?