[OpenAFS-devel] Re: Windows error ERROR_REQ_NOT_ACCEP

[Ben Creech]

After a week or so of logging, I've only gotten one trace of the error. 
This would be because the machines are used much less now that the semester 
is over.

http://www.eos.ncsu.edu/wolfcall/ktc_ForgetAllTokens/ktc_ForgetAllTokens_tr
ace.html

A random hypothesis is that this could be caused by hackers scanning our 
network.  There is a whole slew of people, mostly on EFNet movie/warez 
channels, who have an affinity for 10 mbit up/down for some reason.  That 
problem will be dealt with the heavy-handed way in a week (via port 
blocking at the campus gateway).

Either way, I'd consider it a bug that ambient network traffic could mess 
up the AFS client, unless we're going to put the loopback adapter in the 
installer (and make it "required").  Of course, I have zero evidence that 
this is even the case.  It could be any number of other godawful things 
breaking the AFS Client for us.


--On Sunday, May 11, 2003 1:01 PM -0700 James Peterson <james@abrakus.com> 
wrote:

> The trace option will display the SMB activity that goes on while the
> Cache Manager is active.
>
> The event log information probably won't do you much good without some
> familarity with the SMB protocol.   However, I would be interesting in
> pouring over a copy of the log after this failure occured.   If you set up
> the event log so that it will keep logging (over write the oldest) and
> then when your machine jams, export the event log to a file and email it
> to me. (Not to worry, your password information is not part of the event
> log dump.)
>
> Personally I suspect the error is because the Cache manager is returning
> an incorrect error for a SMB request it doesn't have coding for;
> something like that.
>
> James Peterson
> "Integrity is the Base of Excellence"
>