[jcifs] jCIFS and NetApp with Data ONTAP dynamic home folders

Michael B Allen ioplex at gmail.com
Tue Dec 9 23:45:10 MST 2014


On Wed, Dec 3, 2014 at 11:43 AM, <Marianna.Dorokhina at ricoh-usa.com> wrote:

> Hello  Michael,
>
> The user is trying to deposit a file to his home directory via GSNX
> application (it is our scanning solution).
>
> The GSNX application is using jCIFS API to retrieve client’s home
> directory path from the NetApp (NetApp NAS appliances with ONTAP 7 mode
> versions 8.1.2P4) server.
>
> *Configuration details of the customer environment*:
>
> The customer uses Data ONTAP home directory functionality to create users'
> home directories on the CIFS Vserver and automatically offer each user a
> dynamic share to their home directory without creating an individual CIFS
> share for each user.
>
> (FYI:
> *https://library.netapp.com/ecmdocs/ECMP1196891/html/GUID-741EF656-6B92-4A28-BD98-5942AF22FDA9.html*
> <https://library.netapp.com/ecmdocs/ECMP1196891/html/GUID-741EF656-6B92-4A28-BD98-5942AF22FDA9.html>
> )
>
> The user’s home directory =*\\svuent\fs\us\home_nas\autohome*
>
> The *\\svuent\fs\us\home_nas\autohome* only works if the smb request is
> being executed with the user ID.  The actual DFS link is a number of
> dynamic shares on each of the NAS appliances that host home directories in
> the form of *\\FILERNAME\cifs.homedir*   .
> The cifs.homedir share name doesn’t really exist but it’s dynamically
> created if the username of the SMB request matches a directory name in one
> of the home directories paths configured on the appliance.  This behavior
> is executed over and over again when a user accesses the DFS link until a
> share is presented (this is the multiple target referral portion of DFS).
>
> GSNX is retrieving the home Directory path, but we see
> "STATUS_BAD_NETWORK_NAME” in the network capture.
>
> I wonder if jCIFS API can handle the above home folder model and return a
> valid physical home directory path.
>
> Hi Marianna,

Unforunately DFS in JCIFS is not optimal. For example, if there are
multiple DFS drives that JCIFS needs to try to locate an object, it will
quit after the first one. Technically I don't know if clients are really
expected to retry like that but in practice it has been a problem for users
in the past (DFS volumes with broken replication).

Otherwise, I would really need to study your capture (do not post captures
to the list) to determine definitively what is wrong. If you send me your
capture, I will at least look at it. But even if I see what the problem is,
if it requires a code fix, I would almost certainly just ad it to the TODO
list at this point as I am busy with my own business.

DFS + NetApp is a bad combination for JCIFS.

Mike
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.samba.org/pipermail/jcifs/attachments/20141210/f32eba46/attachment.html>


More information about the jCIFS mailing list