The communication between DB and AP is suddenly broken off

[Former Member]

Dear experts:

We are facing a Network problem between DB server and AP servers.

When Iu2019m logging AP server on, sometimes Network communication error occurs. SAP log(SM21) output says u201CError calling the central lock handler >unable to reach central lock handler.u201D There is no PING responding, but OS is OK when I checked at serveru2019s console. No OS logs shows up.

The problem has happened last September, and we could not found the cause of this problem. All OS patches were applied and changed network switches.

Has anyone faced similar problem, please help us. Any information is welcome.

Our platform is as follows;

・ Windows 2003 x64 SP2

・Oracle 10g (10.2.0.4)

・SAP NetWeaver AS ABAP 7.0 (ECC 6.0) and (SRM5.0)

Our system configuration is as follows

- ERP

・DB x2 (MSCS)

ASCS

CI

Web Dispatcher

・AP x4

DI

-SRM

・DB x1

CI

Web Dispatcher

・AP x2

DI

Any comments are welcomed.

[Former Member]

Let me add some information.

When I turn all DIs off , and working only DB(ASCS+CI). This problem never happens.

All servers we have are same blade serves, Window 2003(64bit), and Oracle 10g.

Regards.

[Former Member]

Eric

Thanks for your response.

-contact your Windows support team and let them handle this with your network team.

We contacted with network team.

They figured out the AP's retransmitting and where to lost packets, but not figured out why the packet delayed to reach Message server. And what is the trigger of not reaching message server.

We have Process Integration (PI) or Master Data Management (MDM). Both of them are single instance, and they never have had network problem although there are same platform and hardware as ERP and SRM (Problem only happened ERP and SRM).

[Former Member]

-Is this 64bit?

Yes, it is 64bit. APs are Standard and DBs are Enterprise.

-run a program such as Ethereal

We installed Wireshark( on AP). Packet is responded from DB, but it takes a few minutes.

So AP tries to retransmit request hundreds of times, then AP node is down.

It happens at all AP servers, but never happens at DB(ASCS).

Edited by: daisuke roppongi on Feb 18, 2010 2:30 PM

[former_member204746]

This is not a SAP BASIS issue, contact your Windows support team and let them handle this with your network team.

[markus_doehr2]

> It happens at all AP servers, but never happens at DB(ASCS).

So I'd check the network cards of the DB server (again) and maybe use different cards (different brand).

Very difficult to suggest something else..

Markus

[Former Member]

Thanks a lot Markus.

-Is this happening randomly ?

Yes, exactly happens randomly.

We built another SAP system, by using same data, same platform(Only hardware is different), and same configutration.

But the problem happed.

Thus, the Hardware(Server and switches) maight not be a cause of this problem.

In addition, When network problem happening, the DI does not show up in TX: SM51.

and PING start responding after SAP instans(DI) off.

[markus_doehr2]

> -Is this happening randomly ?

> Yes, exactly happens randomly.

sigh

> In addition, When network problem happening, the DI does not show up in TX: SM51.

Well - if the messagesserver is not reachable it can't show up.

> and PING start responding after SAP instans(DI) off.

Is this 64bit?

Markus

[Former Member]

Markus

Thanks for your response.

Yes, it is a network problem. We try to figure out the cause of this network problem is server or SAP.

We already changed Network Interface Card, but it did not help.

Nothing on Windows event logs.

Windows firewall is on, but there is no Hardware firewall between clitents.

We have tried Windows firewall off many times, but the situation has not changed.

Any small information is fine, this is critical. Please help.

[markus_doehr2]

> Yes, it is a network problem. We try to figure out the cause of this network problem is server or SAP.

If a ping is not responding then the cause can't be the SAP system. A SAP application such as a workprocess just makes use of the Windows runtime, it does not do anything on a lower area (like network) - it just uses it.

> We have tried Windows firewall off many times, but the situation has not changed.

Is this happening randomly or can you say it's occuring only on specific times? I think about data transfers taking place that are more than normal, a running backup or other things.

I could think of doing some more network monitoring such as putting another machine on the same network segment, configure a promiscuous port on the switch so it receives all network traffic and then run a program such as Ethereal to see if you have a problem with the switch itself (it's software/configuration) or with the network card.

Markus

[markus_doehr2]

> When Iu2019m logging AP server on, sometimes Network communication error occurs. SAP log(SM21) output says u201CError calling the central lock handler >unable to reach central lock handler.u201D There is no PING responding, but OS is OK when I checked at serveru2019s console. No OS logs shows up.

If a ping does not respond you have a network problem. You have any firewalls activated? You see anything in the windows event log? Did you try using a different/new network interface?

Markus

[Former Member]

Moved to SAP on Oracle forum...