Production Broker Monitoring

Has anyone written any tools that monitor things such as internal WM queue-depths, the health of adapters, etc., that they might be willing to share with the group?

We’re looking for monitoring tools that could be used to verify the state of our production brokers, queues, adapters, etc.

Not monitoring BY the broker, but monitoring OF the broker (and its components).

For example, it seems it wouldn’t be too hard to write a program that would use the WM Java API to query the above information, and then write the results to a DB or flat text file, which could be scanned reguarly, with alarms (or e-mails, etc.) sent out, if a bad condition is detected.

Has anyone done anything like this? Or have any thoughts?

Joe

We have developed a webMethods monitoring service, complete with reporting, alerts, and case management (alerts and case management if needed) for approximately 30 different wM variables. These include: # of clients; events queued, published, and delivered; persistent, swap, and guaranteed storage; client subscriptions; service errors; request times; thread usage; service memory usage; etc.

I would be happy to show you a WebEx demo and discuss further. Please let me know.

Laura
650-237-4506
lkoch@inventa.com

I think that the inventa solution is more on the MSP lines.

wM has a tool called the wM manager( a new one) which does all of this.

BMC patrol also has agents that fit in seamlessly if you use BMC to do the enterprise monitoing.

We have done a major study on all the solutions available to us and are not really comfortable about any of them so far.

Shall keep the group posted on this.

Joe,

The webMethods Manager may do all that you wish. See below - took the reqs in your request and provided answers. It is new and you will need to contact your local sales team for release and pricing.

You might also want to search the webMethods site for pages and articles on OMI - Open Management Interface. While you could go down to the older APIs, the OMI may be the better place to start.

Mark Thomsen
Alodar Systems, Inc.

Q: …monitor things such as internal WM queue-depths…
A: Yes. Queue length in docs and in bytes, and queue operations.

Q: …the health of adapters, etc.
A: Yes. The health of running parts as well as some of the logical parts - health of a business process step, of an integration component, of an IS service.

Q: …verify the state of our production brokers, queues, adapters, etc.
A: Yes. The operational status. The current metrics. The connectivity of the components (e.g., which broker an adapter is connected to, which adapter an integration component runs on, which integration component a business process step is implemented with, which step follows another step in a business process, …).

Q: Not monitoring BY the broker, but monitoring OF the broker (and its components).
A: Yes. This is why you would want to go with the OMI and Manager. It monitors the enterprise and B2B components, retaining things like there should be a broker server on a certain host and while it is down and not responding, you still see it, in a down state.

Q: …alarms (or e-mails, etc.) sent out, if a bad condition is detected.
A: Yes. They even built in a web server alerting feature.

The Inventa solution for monitoring can be offered as an MSP or as a fully deployable product solution within your environment (for both the Enterprise Server and the Integration Server.

Our solution also offers a secure web based reporting portal, alert notifications to wireless devices, and reporting on related system variables such as the operating system, database, etc. I am not sure if the wM Manager has that type of functionality.

Fee structures for both options of implementation are very reasonable and can be discussed.

From the demonstration that wM gave us, it sure can do all that you mention.

In addition to that it confirms to the OMI specification that can then tie in to any enterprise monitoring system like Uni Centre TNG or Open view…

My company’s environment is very large and it did cause “issues” for the wM “new” Manager…

Personally, I’m not sold on it… The sales pitch is there and it sounds good, but I think there is some more work needed.

Raymond,

What were some of the issues you experienced?

Joe

While there are more comprehensive products out there, we use the open source netsaint package (http://www.netsaint.org) to help our sysadm team monitor the health of our IS servers. Its quite easy to write Netsaint “plug-ins” to do more detailed status checking. One plugin we use checks the connection from TN to Oracle via Merant every 10 minutes – this was just a small script wrapper around an invocation of /packages/WmTN/bin/tnquery.sh

Hi all,

I’m trying to configure the Monitor to resubmit one service.
That’s what I’ve done till now, and that obviously doesn’t work:

  • activate the audit options in my service

  • In Settings->Extended set the keys :
    watt.server.auditLog = perSvc
    watt.server.auditStore = database

  • create a JDBC Pool and assign it to ISCoreAudit and ProcessAudit. I’ve tested and work fine.

-restart IS

  • execute my service

Afterwards, when I try to monitorize the service, I obtain nothing. I’ve searched for all services but there’s no result.

Does anyone know what I’m doing wrong?

Thanks in advance,

Ignasi

Quick check-List (You may have most of these in place)

Make sure all of you supporting Audit Log/Process tables have been created in the database found in (IS_DIR\common\db\scripts\create)
Also make sure the JDBC Pool user matches the database schema where these tables where created allocates enough connections see Max Connections in your JDBC Pool Alias.

If webMethods has been upgraded, also set the following:

watt.server.auditLoadSchema=true

To test your config, create a service that executes a simple test that can be re-submitted:

Crate a Flow Service that invokes put.math.divideInts
Add Inputs to match num1 and num2 with values 100 and 0 respectively.

In Audit section/Tab set the following:

Enable Audit (when top-level Service Only)
Log on (Error, Success and Start)
Include Pipeline (On Errors Only)

Save and execute … should complain about Dividing by zero.

Go to Server UI … should see the Audit log entry for the above flow service … scroll horizontally to the end and copy the Current ContextID

Go to http://<localhost>:5555/WmMonitor and in the service section perform a search and paste the Current Context Id and do a find.

If the Audit Log is storing and fetching from the same database, it should display your entry.

Repeat the same for your service.