| 1 |
On Fri, Aug 30, 2013 at 12:24 AM, Graham Murray <graham@×××××××××××.uk> wrote: |
| 2 |
> "Stefan G. Weichinger" <lists@×××××.at> writes: |
| 3 |
> |
| 4 |
>> Just found this note from Pacho on planet.gentoo.org: |
| 5 |
>> |
| 6 |
>> http://my.opera.com/pacho/blog/2013/08/27/how-to-write-proper-systemd-unit-files |
| 7 |
>> |
| 8 |
>> I will have to review some of my files then ;-) |
| 9 |
> |
| 10 |
> What I did not understand from reading that is why he (or gentoo policy) |
| 11 |
> does not like 'type=forking'. Reading the systemd man files, I thought |
| 12 |
> that type=forking would be the "natural" choice for most daemons. |
| 13 |
|
| 14 |
On the contrary; with Type=simple systemd has better control on the |
| 15 |
service, since systemd itself execv() the service binary, and it can |
| 16 |
know precisely its PID and when it finishes. With Type=forking systemd |
| 17 |
has to guess what the PID is, and therefore it nees more work to know |
| 18 |
the status of the service. It does a pretty good job, but it's easier |
| 19 |
with Type=simple. |
| 20 |
|
| 21 |
Type=forking is there for old daemons that don't have a --foreground |
| 22 |
or similar option. |
| 23 |
|
| 24 |
Regards. |
| 25 |
-- |
| 26 |
Canek Peláez Valdés |
| 27 |
Posgrado en Ciencia e Ingeniería de la Computación |
| 28 |
Universidad Nacional Autónoma de México |
Archives