Hi Mark, Thank you for quick reply.
On Mon, 1 Sep 2014 10:18:19 Mark Hindley wrote:
> > To my surprise it happened despite "AUTOSTART=0" in
> > "/etc/default/apt-cacher".
> AUTOSTART only refers to the daemon process. apt-cacher-cleanup.pl is run by
> a cron job from /etc/cron.d/apt-cacher
Of course. I'm just wondering whether it make sense to run cleaner when daemon
is disabled?
> > Let alone reason why this process did not terminate within reasonable time
> > I'd>
> > like to highlight the following problems:
> > * cleaner may start multiple times if already running one(s) did not
> > finish.
> That shouldn't happen as there is a lock to ensure only a single running
> process. If this was running on a cache directory where apt-cacher-ng was
> taking locks, I suspect these were not multiple instances, but the various
> forks that apt-cacher-cleanup.pl performs. It is supposed to run quietly
> and at very low priority in the background.
Although I can't support it with hard evidence I'm pretty sure it was multiple
processes, not forks of a one instance. I first noticed this issue with "htop"
which is configured to hide threads by default. Every cleaner process was
nested under "cron". Process tree looked nothing like what I see when I
manually start cleaner. Also "killall" released about 1 GiB of RAM while one
cleaner instance do not use that much memory...
> > * lack of POD documentation
> >
> > (i.e. "perldoc /usr/share/apt-cacher/apt-cacher-cleanup.pl" report
> >
> > "No documentation found").
>
> I accept this, but apt-cacher-cleanup.pl -h does provide something and it is
> also mentioned in apt-cacher(8).
I found it of course but thanks anyway for mentioning.
> I cannot see evidence of a real bug here. My only solution would be to
> prevent simultaneous installation of apt-cacher and apt-cacher-ng.
apt-cacher was not active for several months on this system. I assumed it
would be safe to disable daemon in "/etc/default/apt-cacher". If there is a
conflict with apt-cacher-ng then perhaps it could be considered to introduce
"conflicts" or mention the clash in README.
Anyway since I can't point out the root cause for the problem please feel free
to close this bug if you wish.
Thank you.
--
All the best,
Dmitry Smirnov.
---
What can be asserted without proof can be dismissed without proof.
-- Christopher Hitchens, 2004
signature.asc
Description: This is a digitally signed message part.
