Comment 39 for bug 1569925

@gimpeystrada: Thank you for providing your journal logs (in the future, providing the raw journal file is better, as then we can run `journalctl` on it locally). From the logs:

Apr 26 08:52:13 ICTM1612S02H1 iscsiadm[3234]: iscsiadm: initiator reported error (8 - connection timed out)
Apr 26 08:52:13 ICTM1612S02H1 iscsiadm[3234]: iscsiadm: Could not log into all portals
Apr 26 08:52:13 ICTM1612S02H1 iscsiadm[3234]: :0113,3260] successful.
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Child 3234 belongs to open-iscsi.service
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Main process exited, code=exited, status=8/n/a
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Changed start -> failed
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: Sent message type=signal sender=n/a destination=n/a object=/org/freedesktop/systemd1 interface=org.freedesktop.DBus.Properties member=PropertiesChanged cookie=936 reply_cookie=0 error=n/a
Apr 26 08:52:13 ICTM1612S02H1 systemd-logind[2801]: Got message type=signal sender=:1.2 destination=n/a object=/org/freedesktop/systemd1 interface=org.freedesktop.DBus.Properties member=PropertiesChanged cookie=936 reply_cookie=0 error=n/a
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Job open-iscsi.service/start finished, result=failed
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: Failed to start Login to default iSCSI targets.
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: Sent message type=signal sender=n/a destination=n/a object=/org/freedesktop/systemd1 interface=org.freedesktop.systemd1.Manager member=JobRemoved cookie=937 reply_cookie=0 error=n/a
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Unit entered failed state.
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: Failed with result 'exit-code'.
Apr 26 08:52:13 ICTM1612S02H1 systemd[1]: open-iscsi.service: cgroup is empty

I *think* the issue in this case is that open-iscsi.service is actually in a failed state before shutdown, which I think means it does not run the ExecStop commands? Can you verify that is the case (boot your system, `systemctl is-system-running` should indicate "degraded", `systemctl status open-iscsi.service` should indicate Failed). The problem is that the open-iscsi.service has:

ExecStartPre=/bin/systemctl --quiet is-active iscsid.service
ExecStart=/sbin/iscsiadm -m node --loginall=automatic
ExecStart=/lib/open-iscsi/activate-storage.sh
ExecStop=/lib/open-iscsi/umountiscsi.sh
ExecStop=/bin/sync
ExecStop=/lib/open-iscsi/logout-all.sh

This indicates it, on 'start' of the service, it will try to login to and active all configured iSCSI targets (the two ExecStart lines). However, if either of those fail (as they did in the journal in this case), the ExecStop lines *do not* run. From `man systemd.service`:

           Note that if any of the commands specified in ExecStartPre=,
           ExecStart=, or ExecStartPost= fail (and are not prefixed with "-",
           see above) or time out before the service is fully up, execution
           continues with commands specified in ExecStopPost=, the commands in
           ExecStop= are skipped.

So, I think that we should be using ExecStopPost instead.

@gimpeystrada, can you test locally if your system works correctly by editing /lib/systemd/system/open-iscsi.service to use ExecStopPost rather than ExecStop? You will need to run `systemctl daemon-reload` after editing the service file, I believe.