Coda File System

venus mount of /coda fails

From: Ben Jarvis <bjarvis_at_astro.umn.edu>
Date: Tue, 30 Jan 2001 00:44:39 -0600
Pardon what may be a stupid question but I've been running around in
circles with this for awhile.

I'm trying to setup coda for the first time, I've setup the server
according to the HOWTO and that seems to be working, cmon is able to
talk to codasrv and the appropriate ports are open and daemons are running,
nothing bad sounding in SrvLog

This is just a testing setup so the configuration is almost exactly
what is in the HOWTO as an example, except I've installed under
/usr/soft/coda [and hacked the etc/*.init scripts to reflect this].

When I start venus, it tells me to try again with -init, so when I try
with -init I get:

Date: Tue 01/30/2001

00:33:30 /usr/coda/LOG size is 548341 bytes
00:33:30 /usr/coda/DATA size is 2193364 bytes
00:33:30 Initializing RVM data...
00:33:30 ...done
00:33:30 Loading RVM data
00:33:30 starting VDB scan
00:33:30        1 volume replicas
00:33:30        0 replicated volumes
00:33:30        0 CML entries allocated
00:33:30        0 CML entries on free-list
00:33:30 starting FSDB scan (833, 20000) (25, 75, 4)
00:33:31        0 cache files in table (0 blocks)
00:33:31        833 cache files on free-list
00:33:31 starting HDB scan
00:33:31        0 hdb entries in table
00:33:31        0 hdb entries on free-list
00:33:31 Initial LRDB allocation
00:33:31 Getting Root Volume information...
00:33:31 Venus starting...
00:33:31 CHILD: mount system call failed. Killing parent.

Killed

One thing I notice at the top is that it refers to /usr/coda/LOG and
DATA, which don't exist, the LOG and DATA devices are files in
/home/coda for testing purposes.  I don't think this is the problem,
though.

I also notice in worker.cc out of coda-src/venus that the problem line
causing the above problem is:

      error = mount("coda", venusRoot, "coda",  MS_MGC_VAL,
                    islinux20 ? (void *)&kernDevice : (void
                    *)&mountdata);

error is true so it fails as above.

I'm running linux 2.4.x [2.4.1-pre11 for the moment] with coda support
compiled in.  I've also tried it as a module with the same results.

Initially /coda didn't exist, so it failed to mount.  I tried simply
making the /coda directory and trying again, and now the kernel prints
an error everytime 'venus -init' runs:

coda_read_super: device index: 0
coda_read_super: rootfid is (0x7f000001,0x1,0x1)
Failure of coda_cnode_make for root: error -6

Should /coda exist as a directory or no?

Coda is version 5.3.12.

One possible wrench in the works is that all the filesystems on this
machine are reiserfs.  I'd be willing to revert my root fs back to
ext2 temporarily in the name of science if it would be useful.

Has coda been confirmed to work with 2.4.x?  Has it been confirmed to
work with reiserfs?  

Any help would be appreciated. 

Thanks a lot,
Ben
Received on 2001-01-30 01:45:30