Other than watching server.log or \"pinging\", is

We are Sun Clustering our webMethods v6.5 broker and Integration Servers, and I want to be able to explicitly know the servers are up (finished initializing).

In order to “probe” the IS after it is up and running I am remotely executing a java coded service which simply returns the string “Success”.

My questions are these:

Other than watching server.log for the informational log message ISS.0014.0002C (“Initialization completed in yyy seconds.”), or by continuously “pinging” the server, is there a way to detect an IS has completed initializing?

How do I positively determine if the broker initialization is finished?

Thanks in advance!

John

There are probably several approaches for determining whether the IS and Broker servers are ready;

(a) Create a custom package in IS to perform the following;

  • hava a startup service that could report on the start-up time and broker status
  • have the startup server call the monitoring program or write to the probe logging journal.

(b) Have the probe call a service hosted on IS using a HTTP client or B2B Java client, or Broker Client or whatever other client type you fancy that is supported by wM Fabric and could be used by the probe.

You know another thought is:
For the IS server you will always have a date change for the file
/opt/webMethods6/IntegrationServer/config/server.cnf

and for the Broker you will always have a date change for the file
/var/opt/webmethods/awbrokers65/default/awbroker.pid

Doesn’t tell you if the starts were successful, only would report that the processes exist. I think John’s direction leans toward support answering the question of “Is there an internal service that runs that can be queried to insure that we have a healthy IS and a healthy broker, so that we can do appropriate clustering and failovers.”

Anyone in support out there watching ?? If not, maybe we will file a case and let everyone know what the vendor claims is possible and suggested best practices regarding “health checks/probes”.

  • Randy N.
    Cornell University

After digging more deeply I found…

wm.server:ping - which (if successful) returns what the IS thinks the date and time are

and

broker_ping - which lives in the broker/bin directory and will test the relative health of the broker.

As Randy asks… Will webMethods commit to supporting these commands into the future? I hope so, I’ve built production processing around them.