fleetctl: fleetctl status exits while checking batch of units with one unavailable unit

[kayrus]

Steps to reproduce:

hello@.service:

[Service]
ExecStart=/bin/bash -c "while true; do echo Hello, World %i!; sleep 1; done"

fleetctl start hello@{2..10}.service

$ fleetctl status hello@{1..10}.service
Unit hello@1.service does not exist.
$ echo $?
1

[tixxdz]

Not sure I've to check again, but since I noticed that fleetctl status is a wrapper around systemctl status through ssh, it just executes the command and gets its exit status, so even if the command did execute successfully, the return code from systemctl status can be non-zero! so we just break and that's it.

systemctl status exit code are also not that standard since they may return same exit code for nonexistent units and stopped units. or even if the unit is not on the disk anymore... systemd/systemd#1092

If we can get the output first check some fields, then we may fallback to the exit code, and probably break from that loop only if the error is not related "systemctl status-of-unit", IOW break for all other errors even for systemctl ones...

[jonboulle]

Hm what's the problem here? this seems like working as designed?

[kayrus]

@jonboulle what if I want to print status for all units? current behavior will just stop fleetctl on first failed unit.

[jonboulle]

Oh I see.

I dunno, this has been the behaviour forever. Should we really break it now? People needing granular information can always do a for loop around fleetctl status (which incidentally will perform basically the same since we moved to individual Unit() call(