[Mageia-dev] systemd and non-systemd services question

Thu Sep 27 18:26:28 CEST 2012

'Twas brillig, and Frank Griffin at 27/09/12 16:23 did gyre and gimble:
+> I have a cauldron system with the Oracle Express Edition (oracle-xe) RPM
+> installed.  This has worked nicely in the past, with the simple change
+> that the /etc/init.d/oracle-xe script needs to have LSB headers added.
+> 
+> I don't start the DB automatically at boot, but only when I need it for
+> testing.  Based on the creation date of the files in its log directory,
+> the last time it started was Aug 3.
+> 
+> The database no longer starts.  Running the sysv script produces the
+> message "Starting oracle-xe (via systemctl)", and produces no errors,
+> but nothing happens.  "systemctl list-units | grep oracle" shows the
+> status of oracle-xe.service as "loaded active exited", but no Oracle log
+> file is created.  Moreover, there is no oracle-xe.service file anywhere
+> on my root partition.
+> 
+> Issuing "systemctl start oracle-xe.service" issues no output. Running it
+> under strace shows an execve for /bin/systemctl, and later on an execve
+> for /usr/bin/systemd-tty-ask-password-agent, but very little else, and
+> nothing that even hints at actually trying to execute any oracle code.
+> 
+> It seems like sometime after Aug 3 systemd changed in some way as to
+> cause the failure.  The system was built (and the oracle-xe rpm
+> installed) on Aug 2, and never gets touched after install (and the
+> addition of LSB headers).   I haven't had to start it for a while, so I
+> can't get it more precise than that.
+> 
+> Any idea how to diagnose this ?  Does systemd keep logs anywhere ?
+
+Yeah there are logs everywhere :)
+
+Provided you are root, "systemctl status oracle-xe.service" will show
+you the last few lines of log.
+
+What is interesting here tho', is that it's asking you for a password.
+
+Keep in mind that when running under strace, systemctl will be
+contacting PID 1 and asking it to start the service for you so you won't
+see any direct attempts to to start your service.
+
+You can also look in the journal. If you want all logs related to a
+particular unit you can do: "journalctl _SYSTEMD_UNIT=oracle-xe.service"
+but it might be your issue is actually more related to the process of
+starting it rather than actually starting it.
+
+So in this case you maybe want to just turn up systemd debug logging to
+maximum (see man systemd for the numerous ways to do this, via kernel
+command line, config option or even sending a signal with kill at
+runtime SIGRTMIN+22 is probably what you want).
+
+Then run journalctl -f in one shell and try and start the service in
+another.
+
+Col
+
+-- 
+
+Colin Guthrie
+colin(at)mageia.org
+http://colin.guthr.ie/
+
+Day Job:
+  Tribalogic Limited http://www.tribalogic.net/
+Open Source:
+  Mageia Contributor http://www.mageia.org/
+  PulseAudio Hacker http://www.pulseaudio.org/
+  Trac Hacker http://trac.edgewall.org/
+