CA Catalog does not launch Automic Automation Actions
Article ID: 430425
Updated On:
Products
Issue/Introduction
In our Catalog testing environment we have a Event-Rule-Action configured that performs a "Automic Automation" type action. In the test environment we find this action is triggered fine.
In the other hand, when we configure the same Event-Rule-Action in our Production system, the action is not being launched
Testing has one catalog server node, production has two server nodes behind F5 load balancer.
Environment
Service Catalog v17.x
Cause
Our CA Service Catalog production environment runs multiple application nodes behind an F5 load balancer. Initially, the Event Filter was configured through the public URL behind the F5, which does not guarantee that configuration changes are made on the specific node that will later process the events. Because of this, the rule was actually saved on one node only, while the other node continued running with an older cached version of the configuration. As a result, the rule appeared to "not work," but only when the F5 routed traffic to the node that still had the outdated configuration. This was therefore a cluster synchronization/cache inconsistency issue, not a problem with the rule logic.
Resolution
To resolve the issue, we did the following:
- Accessed each Catalog application node directly, bypassing the F5:
Example: https://node1...
Example: https://node2... - Opened the Event Filter configuration on each node individually.
- Re‑saved the filter on each node (in our case, we edited the description by adding a single character such as a period “.” to force an update).
- After saving the rule separately on both nodes, the configuration was finally consistent across the entire cluster.
After this, the Event Filter started working correctly.
The issue was caused by configuration updates being applied only on one node due to the use of the F5 public URL, which prevented proper synchronization of the rule across all Catalog servers.
Re-applying the rule directly on each application node corrected the inconsistency.
Additional Information
Updates to Rule-Actions are propagated to other nodes using ActiveMQ global messages.
Can you check this page in the documentation to ensure that "Horizontal Clustering" has been implemented? It's more than just installing the 2 Catalog nodes to use the same database:
How to Setup Horizontal Clustering Nodes for CA Service Catalog
Please also check to be sure the following steps in prereqs have been completed.
As ActiveMQ is responsible for message propagation across nodes , then all AMQ nodes should share a common multicast address.
Feedback
