Thanks for your help.  I finally figured out what the problem
was.  Apparently, I needed a /coda directory and this is why the mount
failed.  I had not found that in any of the documentation.  My server
still dies when I try to change the debug level with "volutil
setdebug", as seen below.  This seems weird.  Have you ever used volutil
to change the debugging level, or do you usually set it when you start the
server?

Thanks,
Ryan


On Thu, 10 May 2001, Jan Harkes wrote:

> On Tue, May 08, 2001 at 05:23:37PM -0500, Ryan M. Lefever wrote:
> > I am having problems getting the coda client to work.  I am still
getting
> > the above messages when trying to connect to my own set of 3 servers.  
> > There is nothing in the SrvLog or SrvErr on the SCM from the point
where I
> > try to start venus.  On the server running on the same machine as my
> > client, SrvLog contains this:
> ...
> > [ X(00) : 0000 : 17:04:36 ] E adv_daemon::adv_daemon: AdviceServer    
> > [ A(17) : 0000 : 17:04:36 ] adv_daemon::main()
> > [ H(06) : 0000 : 17:04:36 ] HDBDaemon about to sleep on hdbdaemon_sync
> > [ X(00) : 0000 : 17:04:36 ] CHILD: mount system call failed. Killing
> > parent.
> 
> Just like it says, the mount system call fails. Venus should have
> already made sure that the kernel module is loaded and gotten the
> correct attributes for the coda root directory from the servers.
> 
> Maybe there is a kernel generated message that gives some more info
> about why the mount failed. Check /var/log/kern.log, or
> /var/log/messages, or the output of dmesg.
> 
> > So, I tried to turn the debugging level up on the SCM, and in addition
to
> > killing the SCM, the following messages were output:
> > 
> > --------------------
> > [root_at_nsx /root]# volutil setdebug 100
> > V_BindToServer: binding to host nsx.crhc.uiuc.edu
> > VolSetDebug failed with RPC2_DEAD (F)
> 
> That is unusual, I often use debuglevel 100 on the servers, and it looks
> like it died even before sending back the response. I wonder where the
> server crashed.
> 
> > As a side note, I compiled the server, client, and coda.o from source
and
> > had them installed in "/usr/local/coda".  lwp, rpc2, and rvm were
> > installed from rpms.
> 
> Where did you get the coda.o source from, the latest version is
> distributed as part of the official linux kernels.
> 
> > > > I also tried to connnect to testserver.coda.cs.cmu.edu, and i get
the
> > > > following:
> > > ...
> > > > 18:20:14 Getting Root Volume information...
> > > > 18:20:29 GetRootVolume: can't get root volume name!
> > > > 18:20:59 GetRootVolume: can't get root volume name!
> > > > -------------------------
> > > > 
> > > > As the manual said, I checked /etc/services and it seems
ok.  There is a
> > > > firewall here, and that could be part of the problem connecting to
> > > > testserver, but it should have no bearing on my own servers.
> > > 
> > > Make the following change in /etc/coda/venus.conf
> > > 
> > > - #masquerade=0
> > > + masquerade=1
> > > 
> > 
> > I tried this change, but I still got the same message when I tried to
> > connect to the testserver.
> 
> Ok, your firewall is probably just blocking all UDP traffic. No way of
> getting Coda to work through that. The 'masquerade' change is only
> useful to pierce through a firewall that uses masquerading to hide the
> internal network.
> 
> Jan
> 
>