NetApp filer using NetWorker Snapimage problem

[mpx]

Hi,

I experienced problems with a remote backup of NetApp Filer where NDMP Tape Server is Snapimage installed on a backup server. The environment is following:

- backup server (name: backup1)
- Sun Fire 280 with Solaris 8
- Networker 7.0
- Snapimage 1.6
- Sun StorEdge Tape library L100
- 4x HP LTO 2 Ultrium drives
- 2 drives defined for NetWorker
- 2 drives defined for NDMP/Snapimage
- client (name: filer1)
- NetApp Ontap 6.4
- both computers are connected using 1Gbit ethernet interface, IP communication and name resolving is without problem

Snapimage can correctly work with LTO drives that are defined as NDMP - it can label, mount, umount, read and write tapes. This is OK.
We configured an NDMP client based on the documentation, created a group called filer and ran the group from commend prompt. This is the output of the command:

nsrndmp_save -T dump -s backup1 -c filer1 -g filer -t 1071069684 -l incr -LL -W 78 -N /etc /etc
12/10/03 17:24:23 savegrp: filer1:/etc succeeded.
* filer1:/etc rcmd filer1, user backup: `nsrndmp_save -T dump -s backup1 -c filer1 -g filer -t 1071069684 -l incr -LL -W 78 -N /etc /etc'
* filer1:/etc 'incr' level is not supported for NDMP backups.
* filer1:/etc Requesting Data Server to perform a full (level 0) backup.
* filer1:/etc Connecting to the NDMP data server successfully.
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc data connect: failed to establish connection
* filer1:/etc ******************************************************************
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc Unable to start the NDMP backup process.
* filer1:/etc ******************************************************************
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc Tape server halted: The backup is aborted by operator.
* filer1:/etc ******************************************************************
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc start_backup backup returned with error status, -1
* filer1:/etc ******************************************************************
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc data stop: can't process request at this state
* filer1:/etc ******************************************************************
* filer1:/etc ******************************************************************
* filer1:/etc ------ E R R O R -------
* filer1:/etc Error during NDMP backup
* filer1:/etc ******************************************************************
filer1: /etc level=full, 512 KB 00:00:14 0 file
* filer1:/etc backup failed.

NetApp filer is running ndmp version 4. Ndmpd status is ON. When we run the savegrp command, Snapimage correctly mounts a cartridge. Then it seems that backup starts correctly because it writes 512kb of data to the cartridge but then nsrndmp_save fails.
We tried to set Snapimage in the high debug mode but it generated no output. We also tried to find some Ontap logs on NetApp filer in the /etc/log directory but the ndmplogs are empty.

Could anyone help me with an advice or experience? Thanks in advance.

-- mpx --

 

[605]

Please note your first error message: incr is not a valid level.

Although it seems that the process tries to use a full instead, it could also be that the problem cannot be cured.
Something that you can avoid at least.

Did you define the client as NDMP client?

I think the info about 512kB written is issued before the
process has been executed on the device - it seems to be a
bogus message.

 

[mpx]

Yes. I defined the client as NDMP client. I tried the savegrp with "-l full" to be sure that Full backup will be run. The result is still the same.
I also think that the data written to tape is not real data.

-- mpx --

 

[605]

Make sure that you setup the correct storage node parameter
that it will NOT point to the filer.

 

[mpx]

Storage node parametr is set correctly to "backup1" which is the name of the server where Snapimage is running.

-- mpx --

 

[605]

Make sure that the pools have been setup that way that the
right ones accept the data.

 

[mpx]

I think that the pools are defined correctly. I have defined 'filer1' pool specificaly for this client. It is correctly used by NDMP devices handled by Snapimage. If the backup starts NetWorker correclty mounts a tape from this pool to an NDMP device and then runs the nsrndmp_save command.

-- mpx --

 

[605]

BTW - you did not answer on my very first concern - the level. Incrementals are not allowed !

 

[golddavid]

Snapimage (aka NDMP Tape Server) is an NDMP version 2 product. Try setting ndmp to version 2 on the netapp.

Also, have you taken a look at the netapp ndmp logs? Most often, this type of problem is either ndmp version, or a username/password (remember 8 characters max, and must be root). (Your output doesn't indicate an authentication problem)

Are you sure that your filer has an /etc volume? Normally, you'd use a saveset of filer1:/vol/vol0/etc for that. Try doing filer1:/ to see if that is the issue.

Finally, the 512K written to the tape is simply the first block of data, and only tells you that the beginning of the NDMP process is working; it is only when add'l data is actually written do you know it is working.

Dave