On Tue, May 03, 2005 at 10:19:37AM +0800, Alvin Chan wrote:
> First of all, thanks for your help.
> 
> Now, I have switched to use venus version 6.0.7, the strange errors seems 
> doesn't happen again. But there's still errors when I start venus.
> 
> I check the venus.log and here are the messages:
> 
> [ W(20) : 0000 : 10:14:11 ] Cachefile::SetLength 512
> [ W(20) : 0000 : 10:14:11 ] fsdb::Create: (50224f88.ff000001.fffffffc.5017d048, 62500) failed

This happens when we can't allocate a free object. Also the volume-id is
that of the fake rootvolume (i.e. /coda) so this is sort of unusual.

> And I also find that I can't create any files in the coda directory even 
> after I've clog as an administrator and got the token.
> Could anyone give me more help?

Let me guess...

You start venus, it mounts /coda and then you're trying to copy a whole
bunch of data from some other place into the /coda directory. That won't
work since that directory is a fake directory. All names that are looked
up are being resolved and the client attempts to bind to the servers
that belong to those 'realms'.

If the name doesn't contain '.' characters and such it will probably
skip the lookup and we create a negative entry that indicates that there
definitely isn't a realm with that name. So that's why your other log
had so many realms and filesystem objects in use.

This email tries to describe how it works.
    http://www.coda.cs.cmu.edu/maillists/codalist/codalist-2003/4830.html

Jan