[systemd-devel] Options for logging from service started before journald?

[Lennart Poettering]

On Mon, 19.10.15 18:28, Frederic Crozat (fcrozat at suse.com) wrote:

> Le lundi 19 octobre 2015 à 18:03 +0200, Lennart Poettering a écrit :
> > On Sun, 18.10.15 18:36, Andrei Borzenkov (arvidjaar at gmail.com) wrote:
> > 
> > > 18.10.2015 18:33, killermoehre пишет:
> > > >Am 18.10.2015 um 17:05 schrieb Andrei Borzenkov:
> > > >>What can be done to log from unit that needs to be started before
> > > >>journald? Journal, syslog or kmsg all require journald connection and as
> > > >>far as I understand will deadlock on waiting for journald to accept it.
> > > >>NULL is not an option; is tty the only choice left?
> > > >>
> > > >>Background - openSUSE tries to start haveged before journald to add
> > > >>randomness.
> > > >>
> > > >>This sounds like reincarnation of "how to log to syslog and be started
> > > >>before syslogd". This was solved by moving logging connection to
> > > >>journald and starting it very early. May be something like minilod or
> > > >>blogd that buffers output until journald is started would be appropriate
> > > >>here?
> > > >
> > > >How about not putting haveged in the early part if it can't work
> > > >without? In Arch Linux, haveged is a part of multi-user.target.
> > > >
> > > 
> > > Apparently is some configuration not enough entropy was available. I cannot
> > > decide if journald really needs it or not.
> > > 
> > > http://bugzilla.opensuse.org/show_bug.cgi?id=950857
> > 
> > This bug is nonsense. journald needs no entropy. It used cgrypt for
> > hashing, it never generates any keys with it. As mentioned it needs
> > randomness only for the hashtable seeds, and there it doesn't matter.
> 
> Unfortunately, libgcrypt init does crazy things in FIPS mode (which is
> where the bug is coming from), which needs a lot of entropy (don't ask
> me why :(

Maybe that's the issue to fix then?

Lennart

-- 
Lennart Poettering, Red Hat