On Mon, 16 Nov 1998, Yuxin Ruan wrote:

> Hi,
>   I am in the process of planning for a high-availibility file server,
> and coda seems worth a try. Here are some quiestions: 1) I understand
> that there can be only one SCM per cell. What happens when SCM crash?
> The document says that the cell can function for a brief period of
> time without SCM. How long is this "brief-period-of-time"?(25hr of the
> token lift-span?) What happens if SCM goes down beyond this period?

Nothing bad.  The JCM is just a file server that happens to hold the
"read-write" copy of the shared  databases.


> 2) Is there plan for a quick method of reconstructing SCM in case of
> total failure? 

Yes. As long as you back up a few  databases.

3) By browsing the list, it seems that CODA volumn can
> be exported via SAMBA, what about netatalk? We have a significent
> number of Mac users.  

It would probably require a little hack, when I last tried it, it didn't
work.

4) We have lots of multimedia files been
> shuffled around. These are rountinely over 10MB in size. Would this
> affect the computation for cache size? BTW, the clients will access
> the files only through SAMBA/NETATALK/CVS, so the only coda clients
> will be the two linux servers;) One running samba/netatalk, the other
> cvs. And they act as backup for each other.

I hope that you first play around _quite_ a bit with Coda before
entrusting your data to it.  It's possible to use it, but it's not a
totally robust system.


- Peter -

> 
> Thanks for any reply.
>   Yuxin
>