DataSourcer Redundancy

[Vic]

Hi,

has anyone tried setting up a redundant datasourcers?
I tried it a few minutes ago and noticed several things:
in COLD standby if DataSourcer is killed, backup datasourcer is not started. If both are started, both attempt to access the same ODS and consequently, the backup datasourcer quits because ODS DB cannot be shared.

in WARM standby both Datasourcers connect ot the same DB and consequently backup one shuts down because ODS DB cannot be shared.

in HOT standby Genesys does not support hotstandby in G6.1 for anything but TServer, and there is no G6.5 for Datasourcer. Anyway, I tried, and once again, the same problem happened as with the warm standby scenario.

The error:
Checking Database...
Unknown Message From DB Server :48
Unknown Message From DB Server :48
DB connection #11 opened
The Data Sourcer Application DBID '150' specified in your ODS
is different from DB ID '229' of your application 'Data_Sourcer_Backup'
The ODS must be used for only one ConfigServer and cannot be shared
among multiple ConfigServers.
ODS also may have only one Data Sourcer assigned to it,
and may not be shared among multiple Data Sourcers.
If you are sure that changing ConfigServer is what you want,
you need to delete the record from the table 'OL_DB_PROPERTIES'.

Now, I tried using the latest SCS (6.5.xxxx) and it still did not solve my problem, so the question is:
how do you prevent the loss of data when datasourcer crashes? Since we cannot have a redundant DataSourcer, how would it be possible to correctly collect months worth of data?

What in the world am I doing wrong?

[Alex]

Hi Vic,
as Datasourcer is a Genesys Process that doesn't support Genesys High Availability, we resolve the problem setting up an external H.A. Sysytem (we used HACMP on AIX and Microsoft Clustering on Win2000) and create a resource group only for Data Sourcer and we set a cold standby redundancy that when the first Data Sourcer crashes tries to restart it 3 times and if it is not possible due to any problem, restarts it on the second node of the cluster. It seems it is working well and we only loose a little period of stats, but it is all automatic and it is the only way we found (with genesys engineers) to use H.A. with this component.