MOD_OC4J Does Not Failover Properly After Connection With One Node Is Lost
Last updated on FEBRUARY 16, 2017
Applies to:Oracle HTTP Server - Version: 220.127.116.11 to 10.1.3.4.0 - Release: AS10g to AS10gR3
Information in this document applies to any platform.
Checked for relevance on 12-May-2009
The architecture is a AS 10.1.2 architecture on Linux with with 4 midtiers each one running on its own box:
- 2 OHS-only boxes as frontend
- 2 OC4J-only boxes
- A hardware LBR in front to route requests between both OHSs
- MOD_OC4J load balancing takes place between the OHS boxes to both OC4J instances using Oc4jMount cluster://... syntax.
The problem is load balancing is fine but no dead detection is performed by MOD_OC4J and requests to dead OC4J instances are performed, meaning 50% of the requests can not be served if one instance is down.
The tested scenario consists in unplugging the network cable of one of the OC4J boxes from the network.
The problem can not be reproduced if the OC4J instances are shutdown with opmnctl stopall or if we shutdown the whole box instead of unplugging the cable.
Sign In with your My Oracle Support account
Don't have a My Oracle Support account? Click to get started
My Oracle Support provides customers with access to over a
Million Knowledge Articles and hundreds of Community platforms