SAP R/3 Agent Hanging Issue

[Oliver_Vilwlock]

Hi there,

we are running Veritas BE 10d 5629 with an ADIC i500 hosting 2x LTO3-Drives. We have several SAP Systems from 4.6b up to NW 04s running on Oracle 8.1.7 to 10.2.0 using SAP R3 Agent for Backup.

Unfortunately we often get the following error situation:
If a SAP triggered job starts to run without any free drives it waits until a drive gets ready (which is okay). Although the SAP-job is queued and enters the running state if a drive is getting free it does not start properly, which means that no data is beeing transfered from the agent.

Running the agent in debug-mode leads to the following messages in debug.txt:
07/25/06 12:09:14 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:09:44 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:10:13 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:10:43 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:10:44 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:11:14 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:12:14 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:12:44 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:13:15 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:13:44 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:14:14 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:14:44 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:14:45 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:15:14 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:15:15 VSNTALK: MASTER - Waiting for a vsn msg
07/25/06 12:15:44 VSNTALK: MASTER - Waiting for a vsn msg
......

It seems as if the BE Engine does not send any VSN-Message an therefore the agent does not start transferring data.

Additionally the job "blocks" the free drive, so that all other jobs have to run on the remaining drive until a second SAP-Job "blocks" that one too. This leads to an unsatisfying situation where not backups run at all.

Where is the error and how could it be fixed ? How can we "free" the blocked drive without having to restart the BE-Services (which will then stop the successfull running backups) ?
In version 9.1 we had the same problem with hanging SAP-Jobs but they did not block a drive.

kind regards,
Oliver Villwock

[shweta_rege]

Hello,


Could you please elaborate the exact error or problem ?


What is the exact error that you receive?

Do you receive any other active alerts?


Thank You...

[Oliver_Vilwlock]

Hi shweta,

the problem is :

Running a job from a SAP-System (e.g. from db13) over the Veritas SAP R/3 Agent will create a job "Backup 0xxxxx".
This job is queued until no free drives are available (so far ist is okay).
Once a drive is getting free, the job becomes active (still okay).
But here comes the problem => No data is transfered, which means the Veritas SAP agent waits for a signal (vsn message) from the Job Engine (see debug log) and on the other hand the Job Engine does not send this signal and shows the job as running without counting Bytes, occupying a drive (which is really sad for the other backups which should run).
This situation resides forever (until i restart the BE Services). I can't stop or cancel the job any more and no data is being send from the SAP agent, who is waiting for an answer from BE.

[shweta_rege]

Hello,




-- Are you backing up on a tape?


-- Kindly install the latest Drivers and then check the status?

-- Does the backup hang when you backup on a backup to disk folder?


Thank You.....

[Oliver_Vilwlock]

Hi Shweta,

- yes, i am backing up to tape !
- latest drivers are installed !
- same problem with backup to disk folders !

Like i already mentioned, it seems, that - after waiting a while for the next ressource to be freed for the job - the backup exec media server does not send a "GO"-Signal to the SAP-Agent, which is waiting unendingly for a signal from the media server.

Taking a closer look it also seems, that the job is freezed, because there is no chance to get it stopped or canceled without a restart of the be-services. Additionally the job is displayed as "running" without physically blocking a tape drive (which means that there is no tape in the drive and no loading operation is performed). Nevertheless no other job which is started afterwards can use the "assigned drive".

regards,
Oliver

[Klaus_Kretschme]

Hi Oliver,

we have got the same problem, but it seems that nobody from Veritas(Symantec) is able to (or is interested in) solve this problem. My case 240-167-426 was closed with the comment that we should plan our R/3 jobs so that they always got an empty device.
I am very interested in a solution of that problem.
If you have got a solution please post it here. I'll also do that.

Regards Klaus

[shweta_rege]

Hello,



-- Are you using a Standalone Drive or a Autoloader?

-- Did you try to create a Drive pool and then target the job to the specific Drive pool?


Kindly refer the following Document:


Creating and configuring drive pools in Backup Exec 9.x and 10.x for Windows Servers


http://support.veritas.com/docs/262932


Thank You.

Shweta

[Oliver_Vilwlock]

Hi,

-- Yes we are using ADIC Scalar i500 Autoloader with 2x LTO III drives !
-- I created Partitions with both drives assigned by default ! I am using drive pools already.

regards,
Oliver

[Deepali_Badave]

Hello,

What is the error message in the event logs.

Regards,

[Oliver_Vilwlock]

once again:

ethere is NO error message ! Both sides (Backup Exec Job Engine and SAP Agent) are waiting for the start of the data transfer and nothing happens.

All other jobs afterwards will get an "unable to get idle device" error.

[Oliver_Vilwlock]

Hello anyone out there ?

I am still waiting for this issue. Any new suggestions ?

kind regards,
Oliver

[shweta_rege]

Hello,

Refer the following Document:


http://support.veritas.com/docs/271451



Thank You,

Shweta