systemd startup script

Nagy Ákos nagy.akos at codespring.ro
Fri May 6 13:34:27 UTC 2016


2016. 05. 06. 16:32 keltezéssel, Nagy Ákos írta:
> 2016. 05. 06. 15:34 keltezéssel, Andrew Shadura írta:
>> On 06/05/16 14:25, Nagy Ákos wrote:
>>> 2016. 05. 06. 14:58 keltezéssel, Georges Racinet írta:
>>>> Hi,
>>>>
>>>> On 05/06/2016 01:51 PM, Nagy Ákos wrote:
>>>>> Hi,
>>>>>
>>>>> I wrote a systemd startus script, and test it on Ubuntu 16.04.
>>>>> I try to create a pull request, but i'm not familiar with it.
>>>> I'm just a reader here, but I wonder why you use the forking style and
>>>> not simply run in the foreground and let systemd handle the
>>>> daemonization (as you did with celery)
>>> I don't know exacly the difference between simple and forking...
>>> First I create script for kallithea, and I left with type=forking, but
>>> probably work with type=simple too.
>>> Celeryd work only with type=simple.
>> Forking means the daemon forks into the background (daemonises), so
>> systemd has to track that and find the forked process, which is a bit
>> error-prone. Simple means the daemon doesn't fork but stays in
>> foreground, and systemd manages it and puts it into the background on
>> its own. Simple is better for systemd because it's more reliable;
>> besides, systemd is then able to capture the stderr/stdout and direct it
>> to the system journal automatically.
>>
>> Do we really need to provide a separate unit for celeryd? Is it not
>> possible to start it in some way other than by running paster?
> Maybe with ExecStartPre= but with simple copying the ExecStart from 
> celeryd.service don't work.
> I try kallithea.service with Type=simple, but don't work, only with 
> forking.
>
It work when I remove --daemon:
ExecStart=/srv/kallithea/venv/bin/paster serve /srv/kallithea/my.ini 
--pid-file=/srv/kallithea/kallithea.pid 
--log-file=/srv/kallithea/kallithea.log

-- 
Nagy Ákos
0748219066
SC Softech SRL
Str. Frunzisului, Nr. 29
400664 Cluj



More information about the kallithea-general mailing list