OK, I re-ran vice-setup. Don't ask me why, but at least the server starts now, 
so it's not lack of RAM. Maybe I had something messed up before. Now my 
clients don't connect.

{root_at_morpheus}-{/}# venus

Date: Tue 11/12/2002

14:41:11 /var/lib/coda/LOG size is 2611887 bytes
14:41:11 /var/lib/coda/DATA size is 10447548 bytes
14:41:11 Initializing RVM data...
14:41:11 ...done
14:41:11 Loading RVM data
14:41:11 starting VDB scan
14:41:11        1 volume replicas
14:41:11        0 replicated volumes
14:41:11        0 CML entries allocated
14:41:11        0 CML entries on free-list
14:41:12 starting FSDB scan (4166, 100000) (25, 75, 4)
14:41:12        0 cache files in table (0 blocks)
14:41:12        4166 cache files on free-list
14:41:13 starting HDB scan
14:41:13        0 hdb entries in table
14:41:13        0 hdb entries on free-list
14:41:13 Initial LRDB allocation
14:41:13 Getting Root Volume information...
14:41:14 GetRootVolume: can't get volinfo for root volume (MATRIX)!
14:41:29 GetRootVolume: can't get volinfo for root volume (MATRIX)!
14:41:35 RecovTerminate: clean shutdown

Tailing the server log shows the connect, but I'm clueless as to what's 
actually happening.

14:39:58 client_GetVenusId: got new host 192.168.1.10:3633
14:39:58 Building callback conn.
14:39:58 No idle WriteBack conns, building new one
14:39:58 Writeback message to 192.168.1.10 port 3633 on conn 17d66bfc 
succeeded
14:39:58 VLDB_Lookup: no more records in VLDB
14:40:13 VLDB_Lookup: no more records in VLDB

Is it possible that I've mis-configured my data partition in some way?

coda:/vice/db# volutil getvolumelist
V_BindToServer: binding to host coda
P/vicepa Hcoda T59b8d F57b6a
GetVolumeList finished successfully

I appreciate any and all suggestions.

-- Jon

On Wednesday 06 November 2002 02:46 pm, Jan Harkes wrote:
> On Wed, Nov 06, 2002 at 02:21:19PM -0500, Jon Tyler wrote:
> > As you'll see below, entries on /etc/services and ports are open.
> > `/etc/init.d/codasrv.init start` begins, `startserver` and `codasrv` show
> > up in the process list then mysteriously disappear. OK not mysteriously,
> > RPC2 faults out, but I don't know why.
>
> Hmm, don't see any conflicting ports in use.
>
> > As I mentioned before, this is a small test server (P100, 16MB RAM). Is
> > it possible coda just won't load under something that small?
>
> testserver.coda.cs.cmu.edu is a P90, so that can't be the problem. Only
> 16MB of memory, maybe... It depends a lot on how much RVM you're trying
> to use.
>
> But still, even in that case, I wouldn't expect it to fail while setting
> up the rpc2 listen socket.
>
> Jan