The init script for mt-daapd has references to ${myservice}, which I believe doesn't exist anymore in openrc (correct me if I am wrong). Thus, the code block that checks for multiple configurations will always fall back to the default one. Also, I need to remove "--pidfile ${PIDFILE}" from the start-stop-daemon command within start(), else it will complain about: * start-stop-daemon: did not create a valid pid in `/var/run/mt-daapd.pid' * ERROR: mt-daapd failed to start although process did get started. Reproducible: Always Steps to Reproduce: 1. install mt-daapd and openrc 2. /etc/init.d/mt-daapd start (gives error about the pid) 3. /etc/init.d/mt-daapd.test start (still use the default conf)
We accept patches ;-)
Created attachment 216621 [details, diff] Fixes init script to correctly acquire service name. This patch fixes the init script to work with OpenRC (I tested version 0.6.0) on both single and multiple instances of mt-daapd. I could not recreate the pid error, but observed the pid file being created and destroyed correctly for single and multiple instances of the server (both in baselayout 1 and 2). So I am assuming that the pid error was fixed elsewhere or incorrectly reported. N.B. the patch uses the temporary method discussed in bug #270646 to detect OpenRC.
My notes on attempting to reproduce the PID bug (with no success): I tested with OpenRC 0.6.0. Because there is no package version information on this bug, I checked out the ebuilds which were available on 2008-07-06. This is mt-daapd-0.2.4.1 and mt-daapd-0.2.4.2. I could not reproduce the bug with either package(the pid was successfully created). There are patches in the package that allows the program to correctly create a pid file, however these were submitted a good deal in time before 2008-07-06. I would have to assume that the reporter was using a very old version of the package (even at the time of posting), or this involves a very old bug of OpenRC , which has now been fixed. I would argue there are probably better things to do than emerge very old versions of OpenRC (no longer in current tree) and test them against very old versions of mt-daapd (no longer in current tree) to reproduce this particular bug. Thoughts?
Thanks for the detective work :)
+ 25 Jan 2010; Patrick Lauer <patrick@gentoo.org> files/mt-daapd.init.2: + Fixing init script for #230917 Actually fixing things is more fun!
