Re: NNTPC: problem starting

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

 



[OK, mostly off topic, non linux interested parties skip this now]

On Thu, Apr 09, 1998 at 10:15:43AM -0500, Christopher A. Kalin wrote:
> 
> Alan Brown wrote:
> 
> > On Thu, 9 Apr 1998, Chris Wedgwood wrote:
> > 
> > > Off topic, 2.0.32 is evil. Drop back to 2.0.29 or try alans pre-2.0.34
> > > stuff.
> > 
> > I'll agree with Chris 110% about 2.0.32
> 
> What's evil about it?  Was any evil exorcised in 2.0.33?
> 
> > 2.0.34 is up to pre7 and is behaving well here on my test box, I'm
> > planning to move it to the news server in the next 3-4 days.
> 
> kernel.org doesn't have anything beyond 2.0.33, where could I find it?

There were some fairly substantial changes that occur in 2.0.30 that were
not fully fixed in 2.0.31 or 2.0.32. 2.0.33 is better, but I've still found
2.0.29 to be even more reliable. 

If file size is any indication of the amount of change (which strictly
speaking it isn't):

-rw-r--r--   1 me       users      178161 Feb 12  1997 patch-2.0.28.gz
-rw-r--r--   1 me       users        5123 Feb 12  1997 patch-2.0.29.gz
-rw-r--r--   1 me       users      293692 Apr  9  1997 patch-2.0.30.gz
-rw-r--r--   1 me       users      999854 Oct 18 11:25 patch-2.0.31.gz
-rw-r--r--   1 me       users       41721 Nov 18 15:34 patch-2.0.32.gz
-rw-r--r--   1 me       users       38497 Dec 17 11:55 patch-2.0.33.gz

These changes were made for good reasons adding things like SYN flood
protection and other small network fixes.

The pre-2.0.34 stuff is available from ftp://ftp.uk.linux.org/ (I can't
check it right now, it appears to be down). I have them here if you want I
can send them over.
 
> Sorry for continuing the off-topic stuff, but if kernel evilness is
> affecting NNTPCache, at least there's SOME minor relevance here.

I don't know that you see many funnies specific to nntpcache. Most likely
though it will just trash kernel buffers from time to time, especially if
you get `tear-dropped' or random lockups, especially under high load.
 
> OBTopic:  Count another vote for the AUTH stuff getting put back in.

One more here to. But a nice abstracted way that doesn't rely on a password
file or anything too unix-like. (A specific authorized.users file would be
OK, I could periodically generate that from a dastbase).

I'm going to hack this in myself later this month if nobody beats me to it.
 
> And does anyone have any ideas on how to deal with a Lotus Notes
> newsreader that wants to be in feeder and reader mode at the same time?
> This little (IMHO very broken) trick foxes every newsserver I can point
> NNTPCache at, since INN (for example) wants an incoming connection to be a
> feeder or a reader, but not both.  Can NNTPCache take a IHAVE and emulate
> a POST to the downstream news server, for example?  I imagine that would
> fix this whole problem.
> 
> And no, I can't toss the box out the window, it's a client's.

Tell you client that Notes does not comply with the specs. and special hacks
are required to support it that will cost $. I'm sure there are many people
of this list prepared to hack all sorts of things for the right about of $,
myself included.

Persoanlly, I'm getting rather sick of hacing to find workarounds and fixes
for all the shitty broken MacOS and Windows software out there written by
people who obviously don't pay much attention to the specs, SMTP is by far
the worst.




-cw


[Index of Archives]     [Yosemite]     [Yosemite Campsites]     [Bugtraq]     [Linux]     [Trn]

Powered by Linux