lock method used for buffindexed is now changed
Russ Allbery
rra at stanford.edu
Mon Jan 12 07:41:11 UTC 2004
Przemys=B3aw Maciuszko <sal at agora.pl> writes:
> Compiling it with 16k size wasn't really a good idea...
> Jan 12 08:26:14 inews innd: [ID 493423 news.warning] buffindexed: No magi=
c cookie found for buffindexed 0, initializing
Well, there's the not backard compatible part....
> Jan 12 08:26:14 inews innd: [ID 765236 news.crit] buffindexed: Could not =
extend group.index: Invalid argument
That's really weird. That means that this is failing:
if (ftruncate(GROUPfd, GROUPfilesize(GROUPcount)) < 0) {
syslog(L_FATAL, "%s: Could not extend group.index: %m", LocalLogName);
return false;
}
but I don't understand how it could. The Solaris man page says that
ftruncate failing with that error message means that the length argument
is negative or GROUPfd isn't open for writing, but innd should certainly
have buffindexed open for writing....
--=20
Russ Allbery (rra at stanford.edu) <http://www.eyrie.org/~eagle/>
Please send questions to the list rather than mailing me directly.
<http://www.eyrie.org/~eagle/faqs/questions.html> explains why.
More information about the inn-workers
mailing list