Cassandra driver instrumentation

[mladens]

Merging cassandra driver instrumentation from kamon-io/kamon-cassandra/pull/6

[eugenemiretsky]

Is there a plan to merge this?

[dispalt]

Im interested too!

[osbornk]

What is the status of this? We would love to use this feature.

[lustefaniak]

Current java-driver is 4.6.1, maybe it would be a good idea to make it more obvious from artifact name that this is for older Cassandra driver?

[ivantopo]

I didn't get the chance to look into the actual code behind the new driver yet, only high level review of the APIs and documentation.

The plan I have so far for this PR is to:

• Merge the PR with support for 3.x driver only. Support for 4.x should come soon after.
• Remove the "cluster" tag. It was removed from the 4.x driver and wasn't even accurate on 3.x.
• Rename mentions of "host" to "node". The word node aligns much better with the language on the cassandra documentation.
• Reduce the number of generated Spans. Out of the box, the instrumentation is generating two spans for every executed query: one for the query and one for the actual round trip to cassandra. It makes sense because there might be more than one round trip if a query needs to be retried or speculated on, but we are still getting a fuckton of useless Spans out of the box:

I see a few options for reducing the number of Spans:

• Turn all the executions/retries into marks in the span. This would allow to keep track of all round trips, while still keeping a single span for the query.
• Have a setting to enable/disable tracking of executions/retries.
• Both?

I'll be working on this in the next few days, feedback is welcome!

[ivantopo]

I just applied the changes I mentioned above. Went for a setting to disable tracking of executions (for reducing the number of Spans).

Changes were pushed to this branch in case anyone wants to take a look!