Coda File System

Re: creating rep_volumes without success

From: Jan Harkes <>
Date: Tue, 5 Mar 2002 11:42:32 -0500
On Tue, Mar 05, 2002 at 03:59:14PM -0000, Mohamed Alwakeel wrote:
> But root or any addional volumes dont replicate accross server to client
> and vica versa, there are no errors on the logs and all seems to be o.k.
> After days of learning how coda works I realised the setup scripts were not
> creating all the volume files, after doing an ls in /vice/vol this is all
> that was there :-
>  BigVolumeList  remote

I believe the rest is either in /vice/db, or created whenever volumes
are created.

>  createvol_rep u.server E0000103 /vicepa
> while vice server running but that generated strange error messaga:-
> /usr/local/sbin/createvol_rep: ELF: command not found

Ehh, what command is failing here? createvol_rep is a shell-script, so
you can run it with sh -x createvol_rep to see where it fails.

> V_BindToServer: binding to host mezzo1
> VolDumpVRDB failed with Unknown RPC2 return code 200
> Failed to dump the current VRDB into /vice/db/

Interesting, there is no RPC2 error code 200, maybe it is an error that
came from the server itself, in which case there should be something in

> Also I tried adding new volume no in VSGDB and trying above again and it
> failed again, even if I use an existing volume no it also fails with same
> error.

If you change anything in the /vice/db/servers or /vice/db/VSGDB files,
you need to restart all servers. These files are only read during

> By the way there is some bugs in the src linux rpms, when I try to a rpm
> rebuild it all seems to be o.k. but fails becuase path to init scripts is
> wrong and it cant find them so fails the rpm recompile/build, again this is
> with version 5.3.18 and 5.3.17, I have not tried any other versions. 

I don't know, I build the RPMs on a redhat 5.2 system, which is pretty
old. It seems like anyone who is building from source simply skips the
package it up in an RPM step.

> Last question with the clog command is there anyway to get the tokens to
> stay longer than the default which I think is 2 days ? or is there a way
> just to avoid this and have it configured so you dont have to keep on
> logging into client box every so often to keep on authenticating to server

Default is 25 hours, no they can't stay longer except if you modify the
source, there is no way to avoid it, use a cronjob
    ( cat /etc/coda-password | clog -pipe username )

to keep authenticated to the server, security wise this is identical to
having a kerberos key in /etc.

Received on 2002-03-05 11:43:55