How to Effectively Monitor System Processes

[Richard101]

ServiceNow in our environment is expected to be up 24/7, as is the case for most of you. Also, we have some important jobs that run in the background, like user creation, interfaces, etc. I was wondering how you monitor things like this on a day to day basis. More specifically:

• Do you have any alerts setup for issues with jobs, MID servers, LDAP servers, etc? How do you have these configured?
• Do you have any monitoring on the ECC queue to see if jobs aren't be processed or backing up?
• Do you have metrics on scheduled jobs like average run time, or number of times run on business rules?
• Is there anything else that you watch out for that has saved you by catching it earlier than you normally would have?
• We have log monitoring agents so we can create incidents when our MID servers are down. What are some log entries or patterns that we could set up alerts on to detect the MID server is not behaving normally?

[Tim Deniston]

We set up something to monitor a client's ServiceNow instance using a web service. Basically, a monitoring tool sends ServiceNow a request and ServiceNow replies stating that it received the response. Pretty simple.

ServiceNow automatically monitors your MID server, but you could also set up a monitoring tool to perform the standard monitoring of the MID server, too. CPU, Memory, processes, etc.

Scheduled import jobs should be monitored and send out alerts if there are errors or failures. This is pretty common practice and is very easy to do.

HI also monitors all of your environments and displays stats about downtime, etc. Just log into HI and you should see the graphs for your instance(s).

I don't think I have seen much beyond that.

[Richard101]

Thanks Tim. We aren't worried about our actual instance going down, though the web service health check is a good idea to keep in mind. For the MID server, we had an issue where the service was still listed as running on our Windows server, but the MID server was reported down in the instance? Do you have a business rule configured to create incidents when this happens? Same with the import jobs?

[Tim Deniston]

My current client's process is to send an email when an import fails. That could easily be changed in order to create an incident. I don't think the MID server monitoring automatically creates an incident either. It just creates an event. You could have a script action that watches for that event and creates an incident. See MID Server Heartbeat - ServiceNow Wiki for details on MID server monitoring that happens by default.