Open linuxmail opened 9 years ago
I'm not following this request, could you give me some more detail or a PR to show what you're talking about?
I use Puppet to install a whole Eleasticsearch with 4 nodes (one replication, two quorums) and it takes a lot of time, until the clusterstate becomes green and all services could connect. So I have to take a look, when I can "restart" Kibana service, otherwise the services fails ... With:
[Service] -/etc/sysconfig/kibana Type=simple ExecStart=/opt/kibana/bin/kibana User=kibana Restart=always
I have nothing to do and systemd tries to restart, until Kibana works. Also if the cluster becomes unavailable (red) and later green again ... nothing to do :-)
As I see it: For Debian EnvironmentFile should be /etc/default/kibana :-)
PRs are welcome to make that configurable, I don't think it should be the default though. Having a service continually being restarted can cause a lot of other issues. I haven't seen the issue where kibana needs to be restarted after ES goes red.
I guess your problem can be solved using https://forge.puppet.com/elasticsearch/elasticsearch#connection-validator
hi,
If you have a cluster based setup, it would be nice, if the systemd tries to restart the Kibana service.