[Nix-dev] Will there be a systemd replacement at any time inthe future?

Alexei Robyn shados at shados.net
Tue Sep 2 05:27:55 CEST 2014


Shea & Michael: By default systemd sends stdout & stderr to the journal,
this is controlled by DefaultStandardOutput & DefaultStandardError in
systemd-system.conf. So yes, if these are set to `journal` (or stdout is
set to `journal` and stderr to `inherit`) and you had stdout/stderr
messages which weren't caught then this is a bug. Were the issues you
were having something you could replicate or...?

Michael: The speed issue with the journal is more likely fragmentation;
the journal fragments like nothing else on my systems, unfortunately,
and the slowdown for journal commands is one of the few annoying issues
I've found with systemd in actual usage :(. journalctl should not
otherwise be slow, unless you're not specifying a timerange and are just
seeking forward from boot or other distant past time. On that note, `-e`
should really be default on journalctl, bleh.

Matthias: Ultimately, adding an alternative init system is a non-trivial
amount of work, and thus like most Nix-related things it'll happen as
soon as someone who cares about it (e.g. you?) devotes the time to make
it happen.

- Alexei

On Tue, Sep 2, 2014, at 09:12 AM, Shea Levy wrote:
> FWIW we've also had problems with logging to stdout not being captured.
> Luckily these were all internal apps and we could fix the bug of logging
> to stdout instead of stderr, but IMO it's also a bug that journald
> didn't capture it.
> 
> ~Shea
> 
> On Tue, Sep 02, 2014 at 02:28:34AM +0400, Michael Raskin wrote:
> > >On Tue, Sep 2, 2014 at 12:26 AM, Michael Raskin <7c6f434c at mail.ru> wrote:
> > >>>As a sysadmin, I love systemd and journald. If you want to maintain lots of
> > >>>disparate things and look all over the OS while troubleshooting, good for
> > >>>you, but systemctl and journalctl make life so much easier. Upstart is a
> > >>>very poor substitute.
> > >>
> > >> What are the keys for journalctl to make it work? Because it wants to
> > >> read hundreds of megabytes of data just for showing me last 20 lines of
> > >> some log — way worse than tail.
> > >>
> > >> Also, could you submit a patch that makes systemd log stderr of services
> > >> reliably (stdout too)? We had it with Upstart, but with Systemd it got
> > >> lost and I am not sure how to make it automatic for all services.
> > >>
> > >
> > >Well, what's make it unreliable? I think systemd always logs
> > >stdout/stderr. And it's seen with "journalctl -u NAME" and "systemctl
> > >status NAME".
> > 
> > vsftpd stderr doesn't get logged if the daemon fails due to 
> > misconfiguration.
> > 
> > Logging worked fine with upstart.
> > 
> > >May be I've not logged hundreds of megabytes to systemd yet. Does't
> > >break when traffic is too high?
> > 
> > This is not that much, actually, for a notebook — dmesg adds up very
> > quickly. 
> > 
> > It is all fine on a SSD, but no other my task needs an SSD so I have
> > a HDD. Journald hates notebooks with HDDs.
> > 
> > 
> > 
> > _______________________________________________
> > nix-dev mailing list
> > nix-dev at lists.science.uu.nl
> > http://lists.science.uu.nl/mailman/listinfo/nix-dev
> _______________________________________________
> nix-dev mailing list
> nix-dev at lists.science.uu.nl
> http://lists.science.uu.nl/mailman/listinfo/nix-dev


More information about the nix-dev mailing list