On 9/20/21 21:23, Javier wrote:
> Also I'm wondering if there's a way for s6 services/oneshots to know if
> there's already internet connectivity available (not that the ethernet
> interfaces are up, or that dhcpcd for example is up, but rather that
> there's internet up and running). I believe since last s6 upgrade (last
> week), now freshclam starts too early, and fills the starting console
> with inability to download the DB... See logs:
I've got some good news. I've actually been working on implementing
primitive network detection in s6-scripts. If you've noticed, there's a
network oneshot in your /etc/s6/sv directory right now. It's not very
good though, so I have not made use of it. I'm working on a way to make
this better and to allow users to leverage it and use it as a dependency
for any service they run. My goal is to get it land in core this week.
> About conflicting logging, perhaps I actually can extend those configs
> to ALL services/oneshots, if managed by s6... For example, perhaps I
> should remove:
>
>> log_file "/var/log/mpd.log
>
> From /etc/mpd.conf, and the same for all other services/oneshots... Is
> that right?
By all means, feel free to disable any service-specific logging options
like that. I personally much prefer using s6-log for logging, so
eliminating redundancy is a good thing. As long as the daemon writes its
logging output to stdout and/or stderr, s6-log will catch it. For things
that log to syslog, I recommend installing and enabling syslogd-s6.
There are other syslog daemons out there of course, but syslogd uses
pure s6 tooling to read /dev/log and it integrates perfectly with
s6-log. If you are using an s6 setup, it is far superior to syslog-ng,
and others imo.
More information about the artix-general
mailing list