daemontools: How to find out what is wrong?

Question

I'm exploring daemontools to run some of our applications, which generate some output as well (on both stdout and stderr).

I installed daemontools on a debian box. Every time, I do a svc -u /etc/service/myservice to start the service, I get a [defunct] process. How can I go about debugging this problem.

Following is my /etc/service/myservice/run:

#!/bin/bash
exec myservice -c /etc/myservice/myservice.conf

and /etc/service/myservice/log/run

#!/bin/sh
exec multilog t ./main

Any help would be apppreciated. Thanks.

More information are velcome. For example, what happens when you directly run /etc/service/myservice/run ? — Gregory MOUSSAT, May 08 '12 at 13:31
It runs just fine, spewing out messages on the console (both stderr and stdout). — Lazylabs, May 08 '12 at 17:30
So try to run it with the same user as daemontools will use for it. This can also be a problem of environement: what happen if you run it from cron ? — Gregory MOUSSAT, May 09 '12 at 00:01

score 3 · Accepted Answer · answered May 08 '12 at 13:36

You can catch some errors with ps -aux | grep readproctitle.

Some debugging ideas:

myserver is not in the PATH. Consider specifying the full path to the program.
myserver forks into background: This is explicitly forbidden with services started using daemontools.
Permissions: Double-check the permissions of the directories and of the run script.

score 1 · Answer 2 · answered May 01 '16 at 18:52

I faced the same issue. When nothing else worked, this worked for me :

Find the pid of svscanboot through ps auxf. You will see that it starts svscan which monitors all the services.
kill -9 -pid.

svscanboot will start automatically and so will all services under svscan.

I faced the issue when named pipe of svcan gets corrupt when no disk space is left on machine.

daemontools: How to find out what is wrong?

2 Answers2