I have a TSM 3-way NDMP running against 5.5.28-1 and 5.6.40-43961.
Server Version 5, Release 4, Level 1.2, Command Line Administrative Interface - Version 5, Release 5, Level 0.1 and for another backup-site Server Version 5, Release 5, Level 1.1.
I use NDMP over IP to a storage pool, not direct to tape.

The only thing I noticed was I had to add a static route to direct NDMP-traffic over the correct interface on the mover. Fastpath didn't work as expected.

Setting NDMP and PAX debug - do you see any traffic from the server trying to establish a session on the mover?

actually according to the support matrix heterogeneous 3-way NDMP is supported with both IBM TSM 5.4 and 5.5 for DART 5.5 and 5.6 with the only restriction that VBB isnt supported.

heterogeneous meaning that the data destination (tape) is not a Celerra but the TSM server

for normal 3-way with another data mover as the destination the support matrix says not supported (strangely)

Didnt we have that some time ago here on the forum and issue turned out to be a dual-homed TSM server ?

urle,

i am running TSM 5.4.2 ....i had this running for almost a year now ..but as soon as i upgraded over the weekend, it's not working anymore. When on the server i issue "backup node SERVER_2 /filesystemname mode=full toc=yes" , the command returns immediately with

"ANR1763E BACKUP NODE: Command failed - see previous error messages or view the activity log.
ANS8001I Return code 4".

If i look in activity log there is nothing there. If i try to modify datamover properties in TSM, like changing username that it uses for ndmp it gives me this error

ANR1776I Data mover SERVER_2 has been updated, however a
connection to the data mover failed. Please check the
attributes of the file server specified in the update
command. (SESSION: 6198)

Rainer,

yeah ..we had this discussion here ...didn't we establish that it was not supported by EMC ...is it now ?

http://forums.emc.com/forums/thread.jspa?messageID=484703񶕟

this is my original thread with dual-home TSM issue:

http://forums.emc.com/forums/thread.jspa?messageID=516622񾈎

In this case the only thing that changed was code on datamovers. Unfortunately i did not document static routes on my datamover before the upgrade, but from i can see they are still there. When i set static routes on datamover they are permanent right ? I guess i am back to the drawing board running tcp/ip traces and see what's going on

These are my definitions:

reg node rzsnfs3-urlefs nasadmin domain=SCC.NAS_9 t=nas maxnummp=4
gra auth nasadmin classes=node node=rzsnfs3-urlefs authority=ow
def datam rzsnfs3-urlefs t=nas hladdress=172.21.101.128 lladdress=10000 \
userid=ndmp password=XXX dataformat=celerradump

you are using port 10000 right ?

Did you check the TSM datamover definition (hladdress and lladdress, userid and password) and did you check for your ndmp-user and password in the mover?

i did ..i even reset password on datamover and in TSM. No dice.

yes, things change

looking through old support matrixes it looks like qualification for heterogeneous 3-way NDMP for TSM 5.4 first appeared in the May 2008 support matrix

Are you using "normal" or heterogenrous 3-way ?

heterogeneous 3-way

Aha .. this is interesting. My original account that is used for ndmp was created without -md5 option. It's been working all this time, but i guess in 5.6 DART wants us to use md5 encryption ? I deleted my original ndmp account and created one with -md5 option, as soon as i did that ...voila ..ndmp is working great. I gotta give Ian props again cause he showed me how to use Wireshark to troubleshoot these issues

this is what i saw in Wireshark when i ran trace on my AIX TSM server:

CONNECT_CLIENT_AUTH Reply NDMP Error:ILLEGAL_ARGS_ERR
Error: ILLEGAL_ARGS_ERR (9)

looks like in 5.6 default parameter md5 is set to 1 ..i just checked my cfs14 running 5.5 and default parameter is 0. So there you go

...now as you talk about it - I remember I had to reset the password too but I thought I didn't remember it correctly.