[jcifs] RE: Connecting to DFS Roots

[Daniel Palmer]

> From looking at the captures there are a few things that are 
> strange to me.
> 
> 1) Explorer somehow has advance knowledge that 
> \\trinity.local\TCS is a DFS point and does a > referreal right 
> from the start. If jCIFS tries to tree-connect to this share 
> it just get's BAD_NETWORK_NAME.

OK - I was wondering if the DFS cache would mean this capture was
useless.  I went looking and found dfsutils on the 2k3 Support Tools.
It let me flush the DFS caches and I'll send you a new capture.

>> I've attached the ethereal captures from this test session - and then

>> emulated it using windows explorer.  (Connecting to 
>> \\trinity.local\TCS \\shadrach.trinity.local\TCS and 
>> \\shadrach.trinity.local\TCS-DFS-Root - all of which behaved normally

>> in Explorer).  Apologies for the other junk in the capture - it's a 
>> live network and I didn't have the same issues on my test lan with 
>> smbclient - so I think jCIFS will prob work normally with it too..
:-S
>
> So smbclient works? Can you get a capture of that? If the hosts in
your test environment are different please also get a capture of jcifs
trying the same
> exact thing as smbclient.

Sorry - I meant on my test network I didn't have problems with smbclient
- on my real network I haven't had any luck either.   I just need to get
a linux box up and running that isn't a server - too hard to get
captures !! :)

> The first think I need to know is how the client discovers that the
addressed resource is a DFS point that requires a referral. Then I need
to figure out
> when the client should *not* ask for a referral even though the tree
is reported to be in DFS (e.g. \\shadrach\TCS-DFS-Root).
> 
> Perplexed,
> Mike

Feel free to suggest new methods of testing - I'll happily give them a
go!

Dan
(About to email you some more captures..)