How does Discovery populate the cmdb_ci_db_hbase_instance table?

[HugoFirst]

We run ServiceNow's discovery process.   We scan a couple thousand servers and as a result, we have hundreds of entries in the cmdb_ci_db_hbase_instance table.

That said, there are several servers where hadoop runs, but there are no corresponding entries in the cmdb_ci_db_hbase_instance table.

I'd like to explain why hadoop is missed in some cases, but found in others.   But I am at a loss as to how this actually works.   I'm pretty comfortable with probes and sensors, but this appears to be a pattern.   And I'm not having any luck searching through the patterns.   So I will appreciate any help I can get with this.

[HugoFirst · Accepted solution]

After more research I found where Discovery decides that an entry should be made to the   cmdb_ci_db_hbase_instance table.

The secret is the ci classifier named "HBase Instance".   The classifier has 2 conditions to meet to create an entry:

1. The parameters must contain the   text "-Dhbase.home.dir"

2. The process must listen on one the following ports: 60000, 60020, 16000, 16020

It turns out that the hadoop instances that we not included in our CMDB were listening on other ports.   So we'll need to decide if we want to reconfigure those processes, or update the classifier.

In any event,   I thought I'd include this note in the event that it helps someone else.