[OpenAFS] Re: Windows TGS_REQ on alternate Netbios Names

Rodney M Dyer rmdyer@uncc.edu
Sun, 30 Nov 2003 14:11:33 -0500


At 05:37 PM 11/29/2003, Jeffrey Altman wrote:

>The loopback adapter hack will be removed when the underlying problems it 
>works around have been fixed.

Gee thanks, I guess I didn't think of that (sarcasm).  ;P

What I really mean is what time frame is involved here?  I mean it isn't 
like there's going to be an installable filesystem AFS client soon 
right?  So we are stuck with SMB.  So how are we going to work out AFS 
issues with SMB when most of the issues are caused by the SMB interface in 
the first place?  It is my experience that pulling the net cable out on AFS 
causes it to crash.  This is an SMB interface problem, not an AFS problem 
right?  So the fake SMB server code in the AFS client for Windows wasn't 
written well enough to handle a pulled cable?  Is that the code that needs 
fixing?

Sorry so many questions.  I like to be clear on future development.  I like 
to know about things before I'm impacted by them.  The OpenAFS 'Projects' 
and 'Roadmap" leave much to be desired.

So while I'm here...

We are currently at 1.2.9a and plan on a Jan release of 1.2.10 if 
possible.  I've observed a bad problem with 1.2.9a that I'd like to 
share.  Copying large files >250 Meg to or from AFS causes a severe 
slowdown of our XP machines.  The slowness remains even after the files are 
finished copying.  The only thing that will cure the problem is a shutdown 
and restart of the AFS service.  Our cache is set to 256 Meg so this is 
very coincidental.  I've tried varying threads, services, and chunk sizes 
but nothing helps.

Is this a problem that anyone else has seen?  In the coming week I will be 
testing 1.2.10 to see if it fixes the problem, so I'm curious if it exists 
in that version also.

Ever curious,

Rodney