MSCS - Cant connect to (A)SCS message server!!!

[Former Member]

Hello all,

I've installed an ERP 2005 SR2 HA platform as described on the Installation master guide:

I've installed:

- Clusterized ASCS instance

- Clusterized SCS instance

- Run Sapinst on first MSCS node

- Run Database Instance

- Run Sapinst on second MSCS node

- Installed locally on node A and B, Enqueue Rep Server (4 instances in total)

- Installed locally on node A, Central Instance

- Installed locally on node B, Dialog Instance

Everything worked perfect and Im able to switch the cluster resource group from one node to another... but still I have an issue:

I set up my SAPLogon so I can connect to my ASCS instance. When trying to connect I get "Waiting for connection..." and nothing happens. Not even a timeout. Im not being redirected to the instance on where the cluster resource is currently running.

I did the following tests to troubleshoot this:

- I've increated the trace lavel for the ASCS message server when doing this, and I get a lot of "MS_LOOP" entries, but no error messages.

- I've also tested with niping tool, and I get "Connection to message server o.k." and It stays there without timeout.

- I've increased trace level on SAPLogon, but the same as in previous tests: the client is waiting a response from message server.

More info: the dispatcher on both CI and DI are working perfect as I can connect to them directly and logon to the system.

Any ideas of how i cand further troubleshoot this?

Thanks all for your help!

Joan

[Former Member]

I just went through the same install. I figured out that things are little different (MSCS wise) on the NW04S platform. In the old MSCS installation, the CI is in installed 'inside' the cluster. It looks like with NW04s the cluster works like a Java cluster in that only the SCS and enqueue replication (not 100% sure on the enqueue) are in the cluster, but the CI and any DIs are outside of the cluster and take on different system numbers if installed on a clustered server that hosts the SCS already. In my case, since the SCS was the first install, it took on system number 00 (default), when I installed the CI, I assigned 01 to the CI as it cannot be the same system number (per the directions I think). So, when you connect with SAP gui, you have to go to 01, not 00. The CI install is the piece that created and started the disp+work process, as it didn't exist until this install, as I went and looked for it. So, short answer, try to connect to 01 -- it worked for me.

Our problem is that we are moving to new hardware and our current system uses the old cluster -- no separate CI -- all in the cluster (which is an upgraded system), but the install creates this different layout and our SAP Guis all point to 00 and cannot be easily changed. I assume the answer to my problem is to reinstall the cluster, make the SCS 01 and the CI 00, so that all of our guis can connect. We could also put an application switch in front of everything and manipulate (read spoof) the port from 01 to 00, so that the connection will succeed to the backend.

[Former Member]

Thanks for your reply, Jack.

Unfortunately, Im not able to connect either to 00 nor 01, as I´ve installed my SCSs with both 70 and 71 instance numbers. Too bad...

I made a following work-around:

Since my central instance port number on node A is 76, I´ve reinstalled the Dialog instance on node B with port number 76, so the switch is seamless, as I will connect directly to my CI.

Thanks and regards.

Joan

[Former Member]

in 700 based HA configurations SAPGuis should connect to ABAP Servers using Loadbalancing (logon group based connect).

In this case you have to specify the message server which is running under the virtual hostname of the sap cluster group.

peter