Hi Jan,

Many thanks for the clarification.

Now everything is working fine and I have 3 clients connected to a 
read/write volume.

Thanks very much.

I will try to do server replication in a few days.


Prashant



Jan Harkes wrote:
> On Fri, Jun 08, 2007 at 03:44:06PM +0100, prashant wrote:
>   
>> The issue was with the realms which did not contain the server FQDN and 
>> ipaddress
>>
>> root_at_node2:/coda# more /usr/local/etc/coda/realms
>> node2.bluebarracuda.eu 192.168.1.70
>>
>> I restarted the server and it was working.
>>     
>
> The servers actually don't read the realms file, so the server restart
> must have made the difference.
>
> Are those ip-addresses statically assigned? Coda servers try to resolve
> their own hostname during startup and use the ip address they have at
> that time when a client asks where a volume is located.
>
> Also, clog and venus, which do use the realms file only check
> /etc/coda/realms although the install places it in /usr/local/etc/coda.
>
>   
>> root_at_node2:/coda# createvol_rep volume node2.bluebarracuda.eu
>> and the following was created in the coda directory
>>
>> root_at_node2:/coda# ls
>> node2.bluebarracuda.eu@
>>     
>
> Not related, you created a volume named 'volume' on a single server.
>
>   
>> and ROOTVOLUME in /vice/db/ROOTVOLUME
>>
>> root_at_node2:/vice/db# more ROOTVOLUME
>> /coda
>>     
>
> But the server expects that the root volume is named /coda. So the
> client is unable to mount a root volume named '/coda' at
> /coda/node2.bluebarracuda.eu
>
> You could either change /vice/db/ROOTVOLUME to contain the name of your
> volume (i.e. echo volume > /vice/db/ROOTVOLUME), or you could create an
> additional volume named '/coda' (i.e. createvol_rep /coda node2.blue...)
>
>   
>> root_at_node2:/coda# cfs mkmount node2.bluebarracuda.eu /web/
>> .: File exists
>> root_at_node2:/coda#
>>     
>
> You cannot create your own mountpoints in /coda, it is a readonly volume
> that tries to automount anything you try to look up. So in this case it
> tried to lookup node2.blue... which created a temporary link, and as a
> result it returned EEXISTS.
>
> Besides your volume was named 'volume' so I'm not sure how it would find
> a volume named '/web/' if you didn't create it.
>
>   
>> --- Is this the proper way to do it??
>>
>> Basically i want to have a volume created in /coda - ie /coda/volume1
>> And want all my client to RW access to this volume.
>>     
>
> Create a root volume and make sure the server knows it's name,
>
>     createvol_rep codaroot node2.bluebarracuda.eu
>     echo codaroot > /vice/db/ROOTVOLUME
>
> Now with a client you should be able to do,
>
>     ls -l /coda/node2.bluebarracuda.eu
>
> And that should show an empty directory, if it shows a symlink the realm
> lookup failed and you should copy /usr/local/etc/coda/realms to /etc/coda/
>
> Once you see an empty directory you can use the following to see the
> name and some other information about the volume,
>
>     cfs listvol /coda/node2.bluebarracuda.eu
>
> Or check on which server(s) it is hosted with,
>
>     cfs whereis /coda/node2.bluebarracuda.eu
>
> Then if you want to add additional volumes you create them on the server,
>
>     createvol_rep web node2.bluebarracuda.eu
>     createvol_rep user node2.bluebarracuda.eu
>
> And then if you are authenticated, mount them somewhere in the existing
> volume,
>
>     mkdir /coda/node2.bluebarracuda.eu/home
>     cfs mkm /coda/node2.bluebarracuda.eu/web web
>     cfs mkm /coda/node2.bluebarracuda.eu/home/user user
>
> If everything worked correctly web and home/user will be empty
> directories and cfs listvol and cfs whereis can be used to confirm that
> these directories are part of the new volumes.
>
> Jan
>
>