0 active servers, conversations not released

Hi All,

We have an odd case where an ACI server written in Natural abends because of a call to COBOL due to some LE inconsistency. When it does, the entire IMS region fails. This, in turn does not, understandably, signal to broker that a shutdown is required.
If the CIS is viewed through Control Centre , the instance of this ACI server is not removed and it shows 0 active servers and a number of active conversations. Is there a way of clearing this instance without bouncing the kernel?


EntireX 7.2.1
Natural 3.1.06

Are you able to issue a shutdown from the Broker control centre? Right-click on the server instance line and select the Shutdown option. From SMH, select the server instance (from the left panel, takes some clicking around to find the right one), then select DeRegister from the bottom left panel.

As you have already noticed, the Broker removes a server instance only when the server deregisters. If a server terminates abnormally the Broker does not known that the server is no longer available.
However, the Broker does not keep this server instance forever, there is an associated timeout. If there is no activity from the server in the specified timeframe the Broker will remove the server instance. Check the parameter SERVER-NONACT in the attribute file. You can have a global value for all services, or different values per service.

The same applies also to conversations, here the timeout parameter is CONV-NONACT. And for clients, see CLIENT-NONACT.