On Fri, Mar 31, 2000 at 06:38:48PM +0200, Alan Scheinine wrote:
> Dear Jan,
>    Thank you for your reply.  However, venus always fails
> now that cancelled-out and remade space for volumes.
> It fails on every machine.  I guess I made a mistake when I
> got the error message
>  13:25:02 fatal error -- CommInit: RPC2_Init failed
> but I still get the message when I am very certain there
> are no venus running.

We should see this message whenever venus is unable to bind to the
venus port (udp 2430). In general this means another venus is running.

>    One strange thing is that after it fails in the way shown
> below, I have two venuses running.
> Start with no venus running,
> venus -init &
> [some output from venus]
> 14:16:22 Venus starting...
> 14:16:22 Fatal Signal (11); pid 30190 becoming a zombie...
> while doing setup I see one process "venus -init &" but
> after it fails I see two venus processes.
> 
> In another case I got the message
> 18:18:19 Getting Root Volume information...
> 18:18:19 Venus starting...
> 18:18:39 fatal error -- vsgent::Connect: (0) failed (-2004)

There are 2 processes because at this point venus forks a child to mount
/coda while the parent services the initial upcalls from the kernel. The
failing connect is strange, what is the contents of /vice/db/VSGDB on
the servers?

>   Because it failed on every machine, on one machine I installed
> coda-5.3.4 (I was running coda-5.3.3)
> and starting venus still fails, this time with the message in
> the console file:
> B scan (5400, 129603) (25, 75, 4)
> 18:10:26 fatal error -- CacheFile::ResetContainer: open failed (2)
> 18:10:26 RecovTerminate: clean shutdown
> Assertion failed: 0, file "fso_cachefile.cc", line 132

The layout of /usr/coda/venus.cache changed between 5.3.5 and 5.3.6,
which is probably the cause of this error message.

Jan