slapd does not automatically restart on failure

[Kseniya Blashchuk]

Slapd does not restart after a failure. Slapd package does not include
systemd unit file at all, only init.d script. Makes sense to make a
unitfile with a Restart=on-failure option.

$ lsb_release -rd
Description: Ubuntu 16.04.5 LTS
Release: 16.04

$ apt-cache policy slapd
slapd:
  Installed: 2.4.42+dfsg-2ubuntu3.3
  Candidate: 2.4.42+dfsg-2ubuntu3.5
  Version table:
     2.4.42+dfsg-2ubuntu3.5 500
        500 http://nl.archive.ubuntu.com/ubuntu xenial-updates/main amd64
Packages
 *** 2.4.42+dfsg-2ubuntu3.3 100
        100 /var/lib/dpkg/status
     2.4.42+dfsg-2ubuntu3.2 500
        500 http://security.ubuntu.com/ubuntu xenial-security/main amd64
Packages
     2.4.42+dfsg-2ubuntu3 500
        500 http://nl.archive.ubuntu.com/ubuntu xenial/main amd64 Packages

  affects ubuntu/openldap

[Ryan Tandy (rtandy)]

Hello,

You can configure that yourself with a systemd drop-in file that
modifies the generated unit. A similar one was recently added to set
RemainOnExit=false; see bug 1821343.

I can't comment on whether it's appropriate to add that behaviour in the
package by default.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Kseniya Blashchuk (ksyblast)]

Obviously, it's possible to configure it, however I guess it's better to include it to a package rather than everybody would need to do it manually. In my opinion that would help to avoid service interruptions.

[Robie Basak (racb)]

Also see bug 1821927.

[Andreas Hasenack (ahasenack)]

I think a database-like service having such a restart configuration should be a local decision by each administrator. I fear data could be corrupted, for example, if the crash was caused by a bug, and further restarts could make the problem worse.

What I do agree with is that openldap should migrate to a systemd service, away from sysv.