MongoDB Observability with Honeycomb

Database observability is an especially nuanced topic, and MongoDB is no exception. Honeycomb supports several complementary strategies to give you maximum visibility into your MongoDB deployment.

Capture queries on the wire

To quickly track down database problems, it’s important to understand your query patterns along multiple dimensions:

• What’s the breakdown of reads vs writes, including for each individual collection?
• What’s the query latency distribution?
• Are there queries with excessively large request or response sizes?
• How do changes in query families correlate with code deployments?

Honeycomb’s TCP agent for MongoDB gives you query-level visibility without any impact on database performance. It passively analyzes traffic between your MongoDB clients and servers to help you quickly answer these questions and more.

Run the MongoDB TCP collector.

Collect MongoDB logs

Database logs provide additional insight into bad queries, by surfacing information about query plans, index use, locks held, etc. Our agent parses MongoDB logs and extracts useful information such as:

• The normalized query shape
• Number of documents examined
• Duration of the query execution
• Lock duration.

Run the MongoDB log agent.

Record System Statistics

Some metrics that reflect the high-level health of a MongoDB instance can’t be captured on a per-operation basis. These include:

• Queue size
• Running queries
• CPU utilization
• etc.

Honeycomb provides a helper to capture these statistics for you.

Run the MongoDB stats helper.

Alert on problems

Once you’re collecting MongoDB data, make sure you’re the first to know when something goes wrong. Read more about defining triggers and alerts with Honeycomb.

Open Source

Honeytail and our installers are all open source, Apache 2.0 licensed. Their source can be found on GitHub:

• Honeytail
• Installers