[leafnode-list] Re: Fetchnews process takes very long

[Matthias Andree]

Torsten Bronger schrieb am 2008-10-25:

> Hallöchen!
> 
> Reiner Steib writes:
> 
> > On Sun, Oct 19 2008, Torsten Bronger wrote:
> >
> >> In "top", the expensive process calls itself "fetchnews".  I
> >> think texpire is called only once per day or so.
> >
> > Add some -v's to the fetchnews call to generate more verbose
> > output.  This might give some hints where the time is spent.
> 
> The log file with -vvvv is:
> 
> leafnode 1.11.6: verbosity level is 4, debugmode is 0
> try_lock(timeout=5), fqdn="wilson.homeunix.com"
> news.cis.dfn.de: connecting to port nntp...
> news.cis.dfn.de: connected.
> news.cis.dfn.de: using STAT <message-ID> command.
> news.cis.dfn.de: authenticated as bronger
> news.cis.dfn.de: 0 articles posted.
> news.cis.dfn.de: getting new newsgroups
> news.cis.dfn.de: reading server info from /var/spool/news/leaf.node/news.cis.dfn.de
> oecher.kultur: no new articles
> ...
> news.gmane.org: conversation completed, disconnected.
> wrote active file with 36093 lines
> Started process to update overview data in the background.
> Network activity has finished.
> 
> When I hear the heavy I/O on the hard disk, those last lines are
> already written.  So this "update overview data" seems to be the
> cause.

Yup. You can have that synchronously when adding -w to the command line.

-- 
Matthias Andree