Hi all.
Just figured out that in deb packages, provided by your repo, systemd.service's 'Restart=' option is... inconsistent?
$ systemctl cat elasticsearch | grep ^Restart
<empty>
$ systemctl cat logstash | grep ^Restart
Restart=always
$ systemctl cat kibana | grep ^Restart
Restart=on-failure
RestartSec=3
I've tried to google about it, but can't find anything useful so far. So I decided to post it here.
Are there any reasons to make Restart* settings different?
I guess, Elasticsearch.service can have Restart disabled because of possible complications, if some node in a complex cluster will try to restart by itself (I'm not very qualified with setups like these, but I see the possibility).
But, why does logstash.service has 'Restart=always' instead of 'Restart=on-failure'?
Reference:
systemd.service manual for latest systemd release - systemd.service
update:
One more question. I had Just tested this on a clean VM just to be sure that I'm not wrong and found this:
root@box1:~# systemctl cat elasticsearch | head -n 1
# /lib/systemd/system/elasticsearch.service
root@box1:~# systemctl cat logstash | head -n 1
# /etc/systemd/system/logstash.service
root@box1:~# systemctl cat kibana | head -n 1
# /etc/systemd/system/kibana.service
Placing logstash abn kibana service files directly into /etc/systemd kinda violates systemd principles: systemd.unit
Path Description
...
/etc/systemd/system System units created by the administrator
...
/usr/lib/systemd/system System units installed by the distribution package manager