RFC Communication error when doing initial indexing

Question

Hello Everyone,
I have a quick question, when doing the initial index rollup for a BIA index the job is failing with RFC communication error.
We currently have the setting related to " TREXRfcServer threads" set to "Automatic Changes" in order to avoid the RFC Communication errors when loading big loads in BI system.
We currently are on Revision 49.
I also have referred to OSS Note 1138603.
The issue is when I am rolling up the BIA Index the rollup job is failing with RFC Communication errors.
The Global parameters within BIA are:
BATCHPARA - 2
NUMPROC - 5
PKGSIZE - 10000000
SUBPKGSIZE - 20000
We have two app servers with 52 Dialog processes and 25 Background processes.
Any help would be really appreciated.
Thanks
Dharma.

Answer 1

Removing references to systems and ports
Issue:
When the initial indexing is being carried out for an InfoCube to
create the BIA Index and there is more than one application server in
the system the job is failing with the http error.
Hello Everyone,
Thanks for the input, after the automatic changes was configured for concurrent requests the error is different right now.
The error message is,
"Remote communication failure with partner http://<Removed>:<removed>/indexCellTable"
Overview/Background:
When the indexing job is triggered as a background job the failure happens when the S table index is being filled.
In our system where we are trying to do the stress testing we currently have two app servers and one central instance.
When I limit the RFCGROUP in RSDDTREXADMIN table to one App server in terms of logon group the job finishes fine.
Whereas, when I do not limit the RFCGROUP and all app servers are open for access the background job finishes with the error message
Remote communication failure with partner http://<Removed>:<removed>/indexCellTable.
We have looked into the OSS Note 1102652 and our BASIS team confirmed that they were able to ping from BI App servers into BIA blades and vice versa.
We also confirmed if the local gateway was used for connectivity.
The above issue was not in other systems where we only had one server.
We currently are at Revision 49 and SP 13 in BI.
We have E5345 Clowertown blades (2 x 4) 16GB blades.
Please let us know if you need additional information.
Thanks in advance for the input.
Dharma.
Edited by: Arun Varadarajan on Apr 24, 2009 1:36 AM