Coda File System

Re: Sort of working, but I'm missing something...

From: Gareth Bult <>
Date: Wed, 23 Sep 1998 16:56:36 +0100
Hi Folks, I'm now three days into trying to get CODA working under
LINUX. If I can finish it I promise to write an LINUX user FAQ (sorry
but the present instructions [unless I'm missing something] leave a
little to be desired.)

What I have;
2 Servers running coda (Linux 2.1.122) one on 4.6 (256M,12G)  one on 4.7
Both have auth2, upd* venus and codasvr running quite happily.
I have a servers file with 2 servers in it.
I have a /coda (my coda:root)  mounted on both boxes, one's an SCM the
other points at it.
I have created coda:source and cfs mkm'd it on both boxes, I can also
write in both directories.

a)    createvol_rep coda:source E0000101 /archive/db/coda
        works fine. (as long as you don't add a trailing '/')
b)    cfs mkm /coda/user/source coda:source
        works fine on both boxes
c)    writing into /coda/user/source works fine on each box

No Data moves between boxes - I have independent file systems!

d)    su user
        clog <user>
e)    ctokens
        <I have some>

f)    echo hello > /code/user/source/test
        (system call interrupted)
        [Both boxes!]


o    tarring or cp'ing lots of data converts the root dir to a file
(certainly on 4.6, not verified on 4.7)
o    many cfs commands do not work (connection timed out?)
o    cmon looks great once I figure out what it means
o    cfs wd / cfs wr log quite happily but nothing happens.....


As a matter of interest, things I had major problems with;
a)    /etc/services lost the auth2 entry during messing about, was hours
before I noticed auth2 wasn't running...
b)    there is no step-by-step guide that makes any sense (without a
system that is already set up!)
c)    the chances of having a partition on all systems which is the one
you want to use for coda is slim...
        Took me ages to work out that I had to create a
/archive/coda/vicepa on all systems and not
        use different partitions on different systems.
d)    The source tree doesn't seem to compile too well on a clean rh5 .1
e)    The VSGDB / users.coda / group.coda / pwd2* initpw etc makes great
sense once you work out what the
        **** is going on.....

...Again if I can get to use it I'll gladly write a few setup

Received on 1998-09-23 12:00:32