[LAU] [OT] another systemd thread
raffaele.morelli at gmail.com
Thu Jan 8 06:57:06 UTC 2015
On 08/01/15 at 01:00am, Philipp Überbacher wrote:
> On Wed, 7 Jan 2015 12:49:12 +0100
> Raffaele Morelli <raffaele.morelli at gmail.com> wrote:
> > QED you among the others are claiming about nothing concrete, I ran
> > debian on my servers and desktops, switching to systemd did not
> > caused a single issue. I've read tons of posts like your, nothing
> > concrete about systemd, just repeating over and over again the same
> > four silly things about phantom sys admin being stucked by ... what?
> > Furthermore, those who are stucked by filesystem check at startup are
> > either noobs or completely ignorants which can't use tune2fs
> > correctly.
> Hi Raffaele.
> As far as I know Debian switched only very recently. Ralf is using
> Arch, and so do I. I did not look it up, but I think Arch switched to
> systemd about two years ago, so we have a lot more experience and were
> involuntary early adopters.
> In many respects, in everyday operations, systemd just works, but so
> did sysvinit. It is just that everything works differently now, and if
> you run into problems it can be a pain to figure out what is wrong.
> There is a lot less documentation on the net, and by that I mean stuff
> that helps with troubleshooting common problems, not the systemd
> documentation itself. In addition, systemd is a moving target, so what
> might have worked half a year ago may no longer apply. On top of that,
> many Arch people seem to have adopted a potterian attitude over the
> last few years. You can be sure to be scorned at, you may get a
> solution, but you won't get a proper explanation of the cause of the
> problem or its solution.
> I did have some concrete problems. The journal seems to be quite a
> mess. The only reliable thing about it is that it does not work when you
> need it the most. Years ago I had some issue with system freezes or
> crashes and the logs contained nothing. I reported the bug and they
> changed something so that the log gets written more often. Recently I
> had another system freeze, and guess what I found in the logs? The full
> half hour leading up to the freeze was completely empty. Nothing. Nada.
> Not a single line. How are you supposed to debug something if the logs
> are empty?
"Years ago I had"... just like, "once upon a time"
> It is not that much better with small problems. There was one thing I
> did using a systemd service file, of course according to the information
> from the Arch wiki, because without that I would be completely lost. I
> am rather (but not completely) sure it worked properly, for months.
> Recently I noticed that it did not work as expected, the program that
> was started through the service file crashed reliably when it
> absolutely should not. The only thing in the logs was that the service
> failed because it was killed by systemd. End of information. Completely
> useless. Someone in the Arch forums did help me (without explanation),
> a small modification in the service file solved the issue, but I still
> don't know whether the program changed, systemd's behavior changed or it
> never worked correctly in the first place. All I know is that systemd
> killed the program. Guess how much fun it is to figure out why the heck
> it did that.
What are you talking about? What's this thing you did?
<sarcasm>is it a secret between you and the Arch wiki ? </sarcasm>
> How about another grieve? Have you ever tried to figure out the
> configuration of some service? Now with systemd you no longer have just
> configuration files, no, you have configuration directories. Not just
> one. For every single service you may have multiple files in multiple
> directories with some sort of precedence system that you may be able
> to figure out if you find the correct man page. Have fun figuring out
> the configuration. Arch used to be simple, you could configure most
> system level stuff using two files.
What's the matter with multiple configuration files? Have you ever jumped into eg.
/etc/apache2/ ? or /etc/postgresql/*/main/ ?
> And of course there is the more general stuff, like stupid binary logs
> and that they put everything plus the kitchensink into systemd. I just
> now found out that systemd in the meanwhile contains a somewhat
> cron-like thing called timers that is not quite a cron replacement but
> is probably, directly or indirectly, involved to the system freeze I
> described above. I know far less about my system than ever before. But
> at least the gnome users are happy, I guess. I do not see any noteworthy
There's nothing wrong with binary logs if you can read them.
I don't want to dig into this again but this is just another vague post with very unclear
points or at least they can be placed in the "software development workflow" ie.
coding -> testing -> finding bugs -> fix release
«My mama said
to get things done
You'd better not mess
with Major Tom»
More information about the Linux-audio-user