[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [leafnode-list] could not hand over lockfile to child



Jan Knutar <jknutar@xxxxxx> writes:

> Hi,
>
> I get this error message in syslog, apparently when trying to post an
> article:

No, this is not related to posting in any way. Only there are no further
log/verbose messages after posting.

> Aug 30 14:13:46 polaris fetchnews[15401]: could not hand over lockfile
> to child: No such file or directory, waiting until child is done.

Hum, looks like the lock file went away. Does this happen regularly or
just occasionally, particularly if the machine is under load (swap
activity or very high number of active processes should be the prevalent
causes)? The code is somewhat racy (no real synchronization), the child
might remove the lock before the parent can hand it over...

> Not sure what child fetchnews is talking about there, no other
> leafnode processes are running.

The child is usually created to update the overview data after the
fetch, detaches from the parent, that then exists. This is so the
"fetchnews" command exits as soon as it no longer needs the network, and
can do the rest of its housekeeping offline. This is useful for use in
ip-up scripts that end with "killall pppd" after the fetch is
done. Unless you pay per minute, that's harmless.

> /var/lock/news is owned by news.news and has permissions rwxrwsr-x
>
> Using leafnode 1.9.24.rel

Please run fetchnews under strace or truss supervision to figure what
syscall triggers this error message.

-- 
Matthias Andree

-- 
leafnode-list@xxxxxxxxxxxxxxxxxxxxxxxxxxxx -- mailing list for leafnode
To unsubscribe, send mail with "unsubscribe" in the subject to the list