Today, we’re pleased to announce the release of the new Datadog Agent (dd-agent 4.0.0). Aside from making available new integrations to PostFix and Couchbase, we’ve focused our efforts on collecting more metrics from existing integrations, in order to provide Datadog users with more visibility into their application stack.

Central to this enhanced visibility, is a new JMX collector available within the new version of the Agent. This collector makes it very easy to customize metrics from a number of integrations available in Datadog.

We’d like to also extend a hearty thank-you to the community. The contributions that we received from @jslatts and @dcrosta, as well as the feedback and beta testing support from many of you made this new Agent release possible.

Below are the new metrics we collect, broken down by integration, and what you will now be able to see within your Datadog account.

NGINX

• Connections Creation Rate (nginx.net.conn_opened_per_s): measures the number of new connections created per second.

Creating new connections can be a time-consuming operation and NGINX offers a few options to use long-lived connections with clients. You can use this metric to evaluate the impact when changing these parameters (if you can control for inbound traffic).

Postgres

We have added a number of new metrics to support the latest versions of Postgres (9.2 & 9.3):

• postgres.deadlocks: number of deadlocks detected in the database

• postgres.temp_bytes_per_sec: bytes allocated to temporary files per second. Temporary files are often created when sorting large result sets or hashing large tables. Having high temp_bytes_per_sec may mean that work_mem is too small.

• postgres.temp_files_per_sec: number of temporary files created per second

Source: https://www.postgresql.org/docs/9.2/static/pgstatstatements.html

Cassandra, Tomcat, Solr, ActiveMQ

The pre-4.0 dd-agent Cassandra check relied on nodetool to gather performance metrics. Starting with the new 4.0 version, the check now uses JMX to gather equivalent metrics, on a per column-family basis.

Other JMX-based checks (Tomcat, Solr, ActiveMQ) have been migrated to use a generic way to gather JMX metrics via jmxfetch rather than having idiosyncratic collection methods.

When you deploy the new version of the Datadog Agent, your pre-existing configuration files will be migrated to the new format.

For more information, refer to this documentation: https://docs.datadoghq.com/integrations/java/

HAProxy

The pre-4.0 Datadog Agent HAProxy check was producing a large number of per-server metrics and ignoring aggregated backend metrics. In response to your feedback we have updated the check to gather a lot more aggregated metrics. We now track per-backend metrics about:

• request rates

• sessions

• bandwidth

• errors (connection, request, response)

• warnings (retransmits, redispatches)

• HTTP response codes (e.g. 404, 5xx)

Each backend is a tag in Datadog so that you can easily filter, slice, dice and compare backends.

Couchbase

We have long had memcache support so we are delighted to welcome Couchbase to the family of integrations thanks to the contribution of @jslatts.

On top of cluster and node statistics, the integration can gather per-bucket metrics.

If you’d like to be able to quickly collect, graph, analyze and alarm on these new metrics, as well as the many events and metrics that were previously available, sign up for a free Datadog trial. You will have access to this data immediately after the Agent is installed.