On Wed, Feb 05, 2003 at 08:55:46AM +0000, paul.rogers_at_mowlem.com wrote:
> Many thanks for your reply.  I\'m sorry I upset your SPAMASSASSIN.  As
> you can see the e mail is not spam.  

I've upgraded it and reconfigured some things. Hopefully it will behave
a little better now.

> I am open to any suggestions you might have as to what it is that is
> going wrong, so I can fix it.  Most of the problem appears to be
> caused by
> 
> >PAM: NO_REAL_NAME       (1.3 points)  From: does not include a real name
> 
> Is that because FROM is an e mail address?

I've made the filter a bit less aggressive on that match. Basically it
assumes that 'legitimate email' typically has a from line that looks
like, From: Foo Bar <foobar_at_example.com>, while spam often has just
the email address.

> On to your reply.....
> 
> >It is just a file, mine is in /coda/usr/jaharkes/hoardlist, that way I
> >can get at it from any client.
> 
> What I don\'t understand is how the client knows where to find the
> hoard file and subsequently which files to hoard.

The hoard file is just a description that is not used as is by the
client. The 'hoard' command passes the information in this file to the
Coda client which stores the actual profile in recoverable memory (RVM),
it is for efficiency reasons merged with the existing metadata of all
cached objects.

> I am assuming that if the server goes down, the client must have
> craeted a local cache of the files to continue working.
> How does it know what files to cache (Via the hoardfile presumably)?

Any file that is accessed is automatically cached locally as long as
there is space available. However they are not automatically refreshed
when the server sends an callback. The 'hoarded' set of files is a set
that the user has identified as 'must have' files and the client will
actively refetch them about once every ten minutes whenever they are
outdated, or got bumped out of the cache due to cache pressure.

> Where does it store the local cache?
> Presumably this is /coda, but I\'m not sure.

No, anything you see in /coda is conjured up by the Coda client. The
filedata is cached in container files in /usr/coda/venus.cache/, while
the metadata is stored in recoverable memory, /usr/coda/DATA. The kernel
module redirects reads and writes from the 'virtual' files in /coda to
the underlying container files in /usr/coda/venus.cache/.

Jan