Kibana "inactive (dead)" after an update

[rmbleeker]

Kibana version: 5.6.5

Elasticsearch version: 5.6.5

Server OS version: CentOS Linux release 7.4.1708 (Core)

Original install method (e.g. download page, yum, from source, etc.): yum

Description of the problem including expected versus actual behavior: I run Kibana on a CentOS 7 host with yum-cron enabled for unattended updates. After every update for Kibana, the service has status "inactive (dead)" according to systemctl status kibana and shows the message:

Warning: kibana.service changed on disk. Run 'systemctl daemon-reload' to reload units.

I would expect Kibana to not mess with the systemd service file during an update, or at least properly reload it when it does. Logstash and Elasticsearch don't seem to be affected.

Steps to reproduce:

1. Install Kibana and yum-cron
2. Set apply_updates = yes in /etc/yum/yum-cron.conf
3. Wait until there is an update for Kibana, after which the service will be down and needs to be manually restarted

Provide logs and/or server output (if relevant):

/var/log/yum.log

Dec 07 06:41:25 Updated: logstash.noarch 1:5.6.5-1
Dec 07 06:42:07 Updated: kibana.x86_64 5.6.5-1
Dec 07 06:42:09 Updated: elasticsearch-curator.x86_64 5.4.1-1
Dec 07 06:42:10 Updated: elasticsearch.noarch 5.6.5-1

systemctl status kibana.service

   Active: inactive (dead) since Thu 2017-12-07 06:41:26 CET; 2h 20min ag

[jbudz]

Closed by #82049

Starting in 8.0 Kibana will automatically restart on upgrade. In 7.11+ this can be modified by setting RESTART_ON_UPGRADE in /etc/default/kibana.