known bugs (was [CUWiN-Dev] State of the Network?)

[David Young]

On Sun, Apr 17, 2005 at 06:54:28PM -0500, David Young wrote:
> On Sun, Apr 17, 2005 at 05:17:10PM -0500, Sascha Meinrath wrote:
> > Hi all,
> > 
> > The good news is that I'm writing this from my living room!  The bad news 
> > is that I can't connect directly to the Internet:  currently I ssh into 
> > Nicole's node (169.254.17.141) then to Laura's (?) -- 169.254.99.16 -- and 
> > then to Zach's (169.254.118.200) -- from Zach's I can ssh into my e-mail 
> > account.  It's great that the nodes are now communicating; but any ideas 
> > for why the nodes aren't creating a seamless network?
> 
> Bugs.  This means that the *test*bed is working as planned. :-)

There are a few known problems:

        * We're running Mar 9 software on the testbed.  Since Mar 9,
          some bugs have been fixed.  I do not remember which ones.

        * There is at least one bug in the Mar 9 SPF code.  One SPF bug
          triggers a diagnostic assertion and core dump---hslsd prefers
          to commit seppuku than to live with the dishonor of a fubar'd
          routing table.  I believe I fixed that bug last night (I have
          not checked in the fix).  Of course, the Mar 9 software does
          not contain that fix.

        * There is something wrong with ETX so that it produces metrics
          less than 256 (the nominal minimum).  I am not certain that
          this adversely affects the testbed operation, but it might.
          I have a pretty good lead on the cause of this bug.

I know of a new bug on the trunk: on our indoor testbed, it looks like
ETX dumped core on two nodes as it tried to interpret a scrambled buffer.
I'm going to leave some hslsd running overnight to see if it happens
again.

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933