Seeing: Server throttled Success writing SMstore file -- throttling

mki mki at gblx.net
Sun Jun 4 17:53:34 UTC 2000


The Success part of the message is actually the interpretation of 
strerror(0), where errno is of value 0.  From the looks of it, 
in innd.h, ThrottleIOError is getting called, however since errno
is not set, it returns with the Success message, although in reality
SMstore()/SMseterror should have set errno to be of an equiv value as
a standard errno one.

Joe, you might want to add a line at about line 531 in art.c, that looks
like:

    SMerrno = SMERR_NOERROR;
    result = SMstore(arth);
    if (result.type == TOKEN_EMPTY) {
!       if (SMerrno != SMERR_NOERROR) {
+           syslog(LOG_NOTICE, "SMerrno had value %d", SMerrno);
            IOError("SMstore", SMerrno);
+   }
        DISPOSE(artbuff);
        return result;
    }

note that this is not a traditional diff; i just manually did the above
to give you an idea of what needs to be done.  This should give you
an idea of what the exact error was before throttling.  

#define IOError(WHEN, e)        \
        if (--ErrorCount <= 0 || (e) == ENOSPC) ThrottleIOError(WHEN); else

note in the above macro, the error message 'e' is not even passed on
to ThrottleIOError and is merely used to evaluate if the throttle should
occur right away. 

katsuhiro: check out SMseterror and the above macro.

hope this helps

regards
mohan

On Sun, Jun 04, 2000 at 09:15:50AM -0700, Joe St Sauver wrote:
> >Date: Sun, 04 Jun 2000 23:20:25 +0900 (JST)
> >From: Katsuhiro Kondou <kondou at nec.co.jp>
> >Subject: Re: Seeing: Server throttled Success writing SMstore file -- throttling
> >To: inn-workers at isc.org
> >
> >} Server throttled Success writing SMstore file -- throttling
> >} 
> >} Anybody see this before?
> >
> >I haven't seen it, but which storage method are you using?
> 
> CNFS exclusively; storage.conf looks like:
> 
> method cnfs {
>         newsgroups: *
>         class: 1
>         size: 0,1000000
>         options: SPOOL
>         }
> 
> What strikes me as particularly odd is that this is a "Success" report
> rather than a failure report -- if the server was throttled because of a 
> problem, that I understand; what surprises me is to see it throttle as a 
> result of a *successful* operation! :-;
> 
> Joe
> 

-- 
mki at primenet.com          Internet Systems Administration
Tel: 602 416 6400         Global Crossing Telecommunications, Inc.
Fax: 602 744 7354         http://www.globalcrossing.com/
PGP Fingerprint:  21 3C DC 04 14 32 5F E9  67 4E 26 1F 28 0B 0C B0



More information about the inn-workers mailing list