cancel
Showing results for 
Search instead for 
Did you mean: 

PI 7.1 HA

0 Kudos

We are trying to install a PI 7.1 System on Windows/Oracle in HA with MSCS.

The architecture is composed by 2 phisical node.

We clustered Oracle Database, SCS and ASCS.

Locally on the node are installed:

Node 1:

ER ABAP

ER JAVA

DIALOG (D20)

Node 2:

ER ABAP

ER JAVA

CENTRAL APPLICATION (DVEBMGS20)

We installaed on the same cluster a Web Dispatcher.

We use note 1052984 but the clustering doesn't work well.

We have this problem:

If node DVEBMGS20 is down the PI doesn't work. Integration Engine and Business Process Engine doesn't work.

If we start the system (only D20) the WEBDISPATCHER hang we need to shutdown and restart it to allow the connection.

When only D20 is up the PI tools are availble but PI doesn't work.

Have anyone any info about:

1) cluster configuration of PI (is right to consider the solution an HA solution where the system work if D20 or DVEBMGS20 is the only instance up?)

2) point of configuration where virtual host name are user

3) there are any configuration point where the phisical host appear?

Regards

Giuseppe

Accepted Solutions (0)

Answers (1)

Answers (1)

Former Member
0 Kudos

Hi Giuseppe,

From NW 7.0 onwards, you can group only the SCS and ASCS as switchover units along with Enqueue replication.

When your DVEBMGS20 i.e. Node 2 goes down..switch SCS and ASCS to Node1 where you have dailog and database.

This way your dialog instance is connected to central services.So your PI is up.

When D02 fails where the database also resides, only switch over database to Node2. Now you have your central instance connecting to central services and database. So your PI is up again.

So...from NW 7.0 onwards central instance and dailog instance are same and central instance is almost absolete in HA scenarios.

Cheers...,

Raghu

0 Kudos

The behaviour of cluster starting from NW 7.0 is clear.

The problem is the not working PI functionality if the DVEBMGS20 is not up.

SLD is on ASCS/SCS or is active only on one node?

We dind't find clear documentation of what we must accept as normal in PI HA scenario.

We consider that the 2 host are equivalent for every functionality without restriction (JSPM, SLD, Integration Server, etc..).

Regards

Giuseppe