5.5.2

5.5.2 / 10-26-2015

Details

5.5.1...5.5.2

[WARNING] Datadog Agent not reporting metrics after Daylight Saving Time (DST) ends

This release fixes a bug on servers that are configured in local time instead of UTC Time. If your server's clock is configured to use Daylight Saving Time (DST), the Datadog Agent might stop sending metrics for up to one hour when the Daylight Saving Time ends or until it is restarted after the Daylight Saving Time ends.

We highly recommend to upgrade to this version if your server is configured in local time.

Changes

• [BUGFIX] Consul: Send the health state service checks of all nodes. See #1900 (Thanks @jgmchan)
• [BUGFIX] Core: Use utcnow instead of now to avoid the forwarder to run into a locked state. See #2000
• [BUGFIX] Fix pycurl dependency issue with Windows Datadog Agent 64-bit MSI Installer.

5.5.1

5.5.1 / 09-23-2015

Details

5.5.0...5.5.1

Changes

• [BUGFIX] Core: Fix dd-agent command-line interface on Linux. See #49, #51
• [BUGFIX] Docker: Fix Amazon EC2 Container Service (ECS) tags collection. See #1932
• [BUGFIX] Docker: Improve parsing of the cpuacct field and of the container ID. See #1940 (Thanks @joshk0)
• [BUGFIX] HTTP Check: Fix SSL certificate check when specifying a non-default port in the URL. See #1923 (Thanks @dmulter)
• [BUGFIX] Nginx: Fix 'application/json' content_type support. See #1943
• [OTHER] Windows: Ship latest version of Gohai with Windows MSI Installer.

5.5.0

5.5.0 / 09-17-2015

Details

5.4.7...5.5.0

New integration(s)

• Consul

Updated integrations

• Agent Metrics
• Amazon EC2
• Btrfs
• Couchbase
• CouchDB
• Disk
• Docker
• Elasticsearch
• etcd
• Google Compute Engine
• HTTP Check
• JMXFetch
• Mesos
• MongoDB
• MySQL
• Network
• Nginx
• PgBouncer
• PostgreSQL
• Process
• RabbitMQ
• Redis
• Supervisor
• System
• Unix
• Windows Event Viewer
• WMI

Consul check

New Consul check.

Supported metrics:

• Number of Consul Agents in the Cluter
  consul.peers: tagged by consul_datacenter and mode (leader | follower)
• Consul Catalog Nodes Up by Service
  consul.catalog.nodes_up: tagged by consul_datacenter and consul_service_id
• Consul Catalog Services Up by Node
  consul.catalog.services_up: tagged by consul_datacenter and consul_node_id

Supported events:

• consul.new_leader events when a leader change is detected.

See #1628

New Docker check

Datadog agent 5.5.0 introduces a new Docker check: 'docker_daemon'.

In terms of features, it adds:

• Support for TLS connections to the daemon
• New metrics:
  • Network metrics
  • Memory limits
  • Container size (rootfs)
  • Image size
• Support for labels (convert them into tags). Off by default, uses a list of labels that should be converted.
• Support for ECS tags: task name and task version

Backward incompatible changes:

• docker.disk.size metric is renamed to docker.container.size_rw
• Old optional metrics: https://github.com/DataDog/dd-agent/blob/5.4.x/checks.d/docker.py#L29-L38 Are not collected anymore
• Old tags are not supported anymore (e.g. name instead of container_name)

As a consequence, the previous check 'Docker' is now deprecated and will not receive further support.

See #1908

Windows 64bit - Datadog Agent

The Datadog Agent is now available in a 64bit version on Windows.
For more information, please visit our Integrations/Agent page.

Flare on Windows

Datadog Agent flare feature makes easy to ship a tarball with logs and configurations to ease agent troubleshooting. Previously exclusive to Linux, it's now available on Windows.
For more information, please visit our wiki page.

[WARNING] JMX host tag issues & potential backward incompatibilities issues with service check monitors

JMX related checks -c.f. list below- were illegitimately submitting service checks tagged with the host value defined in the YAML configuration file. As it was overriding the agent hostname, with a value often equals to localhost, it was difficult to define and scope monitors based on these service checks.

The issue is addressed. JMX service checks have a new jmx_server tag which contains the YAML configuration host value so it does not replace the actual agent hostname in the host tag anymore.

Warning: these changes affect your JMX-service-checks related existing monitors scoped with the host tag. For more information, please get in touch with support@datadoghq.com

JMX related checks: ActiveMQ, Cassandra, JMX, Solr, Tomcat.

See #66

Deprecation notice

datadog.conf disk options

Disk options in datadog.conf file are being deprecated to promote the new Disk check introduced in the 5.4.0 release. It will be removed in a future version of the Datadog Agent.
Please consider conf.d/disk.yaml instead to configure it.

See #1758

Generic Mesosphere check

The previous generic Mesosphere check is deprecated, in favor of the Mesosphere master and slave specific checks introduced in the 5.4.0 release. It will be removed in a future version of the Datadog Agent.

See #1535

Previous Docker check

The previous Docker check is deprecated, in favor of the new one introduced in the 5.5.0 release. It will be removed in a future version of the Datadog Agent.

See #1908

Changes

• [FEATURE] Consul: New check reporting cluster, service and node wide metrics and events for leader election. See #1628
• [FEATURE] CouchDB: Allow blacklisting of specific databases. See #1760
• [FEATURE] Docker: New Docker check. See #1908
• [FEATURE] Elasticsearch: Collect common JVM metrics. See #1865
• [FEATURE] Elasticsearch: Collect primary shard statistic metrics. See #1875
• [FEATURE] etcd: SSL support. See #1745 (Thanks @KnownSubset)
• [FEATURE] Flare: Add JMXFetch-specific information. See #1726
• [FEATURE] Flare: Log permissions on collected files. See #1767
• [FEATURE] Flare: Windows support. See #1773
• [FEATURE] HTTP Check: Add SSL certificate configuration and validation options. See #1720
• [FEATURE] JMXFetch: Memory saving by limiting MBeans queries to certain scopes. See #63
• [FEATURE] JMXFetch: Wildcard support on domains and bean names. See #57
• [FEATURE] MongoDB: Collect active client connections metrics. Enhance connections, dbStats, mem and rpl metric coverage. See #1798
• [FEATURE] MongoDB: Make timeout configurable and increase the default. See #1823 (Thanks @benmccann)
• [FEATURE] MySQL: Custom query metrics. See #1793 (Thanks @obi11235)
• [FEATURE] Nginx: Option to disable SSL validation. See #1626 #1782
• [FEATURE] PostgreSQL: SSL support. See #1696 (Thanks @bdotdub)
• [FEATURE] PostgreSQL: Support for relation schemas. See #1771
• [FEATURE] RabbitMQ: Collect the number of RabbitMQ partitions per node. See #1715 (Thanks @ulich)
• [FEATURE] Supervisor: Option to select processes to monitor by regex name match. See #1747 (Thanks @ckrough)
• [FEATURE] System: Collect %guest CPU time. See #1845
• [IMPROVEMENT] Agent Metrics: Move stats log's level to DEBUG. See #1885
• [IMPROVEMENT] Core: Log collector runs's exceptions. See #1888
• [IMPROVEMENT] CouchDB: Fail gracefully when one or more individual databases are not readable by the configured user. See #1760
• [IMPROVEMENT] Docker: Add an image_repository tag to the docker check. See #1691
• [IMPROVEMENT] Windows Event Viewer: Better configuration YAML example file. See #1734
• [IMPROVEMENT] Windows: Add Datadog agent version to MSI description. See #1878
• [BUGFIX] Agent Metrics: Fix the configuration YAML example file rights. See #1725
• [BUGFIX] Amazon EC2: Update metadata endpoint list to avoid redirections. See #1750 (Thanks @dspangen)
• [BUGFIX] Btrfs: Track usage based on used bytes instead of free bytes. See #1839 (Thanks @pbitty)
• [BUGFIX] Couchbase: Send service check tags on OK status. See #1722 #1776
• [BUGFIX] Docker: Fallback when Docker Remote API /events returns an invalid JSON. See #1757
• [BUGFIX] Docker: Kubernetes support -new cgroups path-. See #1759
• [BUGFIX] Docker: Strip newlines from API responses to avoid parsing issues. See #1727
• [BUGFIX] Google Compute Engine: Update hostname to be unique. See #1736, #1737
• [BUGFIX] HTTP Check: Handle requests timeout exceptions to send the appropriate service check. See #1761
• [BUGFIX] JMXFetch: Do not override service checks's host tag with JMX host. See #66
• [BUGFIX] JMXFetch: Do not send service check warnings on metric limit violation. See [#73...

5.5.0-rc.1

Merge pull request #1891 from DataDog/yann/redis-switch-metric-types2

[redis] `keyspace_hits` metrics: 'gauge' → 'rate'

5.4.6

5.4.6 / 09-08-2015

Details

5.4.5...5.4.6

Changes

[BUGFIX] Disk: Force CDROM (iso9660) exclusion. See #1786
[BUGFIX] Disk: Recalculate disk.in_use to make consistent with df's 'Use% metric'. See #1785
[BUGFIX] Gohai: Improve signal handling for df timeout. See #16
[BUGFIX] Process: Correctly handle disappearing PID. See #1721 #1772

5.4.4

5.4.4 / 08-05-2015

Source Install only

Details

5.4.3...5.4.4

Changes

• [BUGFIX] Update ntplib to 0.3.3 to fix source install requirements. See #1792

5.4.3

5.4.3 / 07-13-2015

Linux or Source Install only

Details

5.4.2...5.4.3

Changes

• [SECURITY] The deb and rpm packages now bundle OpenSSL 1.0.1p. For more details, see the security advisory.
• [BUGFIX] Docker: Do not fail when the event API returns a bad JSON response. See #1757

5.4.2

5.4.2 / 06-18-2015

Linux or Source Install only

Details

5.4.1...5.4.2

Changes

• [BUGFIX] Disk: Strict backward compatibility with old disk check. See #1710
• [BUGFIX] Etcd: Do not query leader endpoint on followers (was causing check failure). See #1709

5.4.1

5.4.1 / 06-16-2015

Linux or Source Install only

Details

5.4.0...5.4.1

Changes

• [BUGFIX] Disk: Get metrics only from physical disk by default. See #1700
• [BUGFIX] Kafka: Fix indentation issue in the configuration YAML example file. See #1701

5.4.0

5.4.0 / 06-16-2015

Details

5.3.2...5.4.0

New integrations

• Mesosphere master
• Mesosphere slave

Updated integrations

• Disk
• Docker
• Elasticsearch
• etcd
• Fluentd
• HAProxy
• HTTP Check
• JMXFetch
• Kafka consumer
• Mesosphere
• MySQL
• NTP
• PHP-FPM
• PostgreSQL
• Process
• SQL Server
• System
• TCP Check
• Varnish
• WMI

Agent Developer mode

The Agent Developer Mode allows the user to collect a wide array of metrics concerning the performance of the agent itself. It provides visibility into bottlenecks when writing an AgentCheck and when making changes to the collector core.
For more information, see our wiki page.

Deprecation notice

Old agent commands

Old command line tools dd-agent, dd-forwarder and dogstatsd are deprecated as for 5.4.0. dd-forwarder & dogstatsd will be removed in a future version, and dd-agent functions will be limited to check, jmx and flare.
For more information, see our wiki page
See #1457, #1569

Ganglia integration

The Ganglia integration is deprecated and will be removed in a future version of the Datadog Agent.
See #1621

Changes

• [FEATURE] Agent developer mode. See #1577
• [FEATURE] Application names to tags ("dd_check:appname") support. See #1570
• [FEATURE] Service metadata support. See #1611
• [FEATURE] Dogstream: Add filename globing support. See #1550 (Thanks @gtaylor)
• [FEATURE] Elasticsearch: Add pending tasks metrics. See #1507
• [FEATURE] Elasticsearch: Add tags to events. See #1444
• [FEATURE] etcd: Add etcd latency metrics. See #1429
• [FEATURE] Flare: Add commands standard error stream to content. See #1586
• [FEATURE] Fluentd: Add type tag support. See #1623 (Thanks @yyamano)
• [FEATURE] HAProxy: Add new time metrics available in 1.5. See #1579 (Thanks @warnerpr-cyan)
• [FEATURE] HTTP/TCP Check: Add support for custom tags. See #1642
• [FEATURE] JMXFetch: Add service check count to JMX checks statuses. See #1559
• [FEATURE] Mesosphere: New checks for masters -reporting metrics from leaders- and slaves -reporting metrics from the selected tasks-. See #1535
• [FEATURE] MySQL: Add threads running metrics. See #1613 (Thanks @polynomial)
• [FEATURE] PHP-FPM: Add custom ping reply support. See #1582 (Thanks @squaresurf)
• [FEATURE] System: Add system swap metrics. See #1549
• [IMPROVEMENT] Limit process restart attempts on Windows on a specific time frame. See #1664
• [IMPROVEMENT] Only start the Collector and Dogstatsd when needed. See #1569
• [IMPROVEMENT] Use internal /run for temporary pid, pickle and JMXFetch files. See #1569, #1679
• [IMPROVEMENT] Disk: New check based on psutil replaces the old system check. See #1596
• [IMPROVEMENT] JMXFetch: Run JMXFetch as dd-agent user. See #1619
• [IMPROVEMENT] NTP: Use Datadog NTP pool. See #1618
• [IMPROVEMENT] Process: Cache AccessDenied failures and PID list. See #1595
• [IMPROVEMENT] SQL Server: Set a timeout for each SQL command, default to 30s. See #1446
• [BUGFIX] Cast service checks messages to strings. See #1617
• [BUGFIX] Fix incorrect EC2 metadata resulting from the no proxy environment settings. See #1650 #1594
• [BUGFIX] Uses NTP check settings to avoid failures in status checks. See #1651 #1558
• [BUGFIX] Disk: Fix a bug where all devices were ignored if the device blacklist regex was empty. See #1666
• [BUGFIX] Docker: Fix intermittent failures (bad support) when a container has no name. See #1565
• [BUGFIX] Elasticsearch: Fix data being illegitimately filtered out when the local node reports under a different hostname. See #1657
• [BUGFIX] HTTP Check: Fix status type errors in service check. See #1644
• [BUGFIX] JMXFetch: Clean previous JMX status python file at start. See #1655
• [BUGFIX] JMXFetch: Fix jmx agent commands false warning. See #1612.
• [BUGFIX] JMXFetch: Format service check names prefix names to strip non alphabetic characters.
• [BUGFIX] JMXFetch: Properly exit on Windows when a specified file is created. See #1643
• [BUGFIX] JMXFetch: Rename 'host' bean parameter to 'bean_host' in tags to avoid conflicts.
• [BUGFIX] JMXFetch: Support bean names that have an attribute with an empty value.
• [BUGFIX] Kafka consumer: Add timeout for ZooKeeper and Kafka connections. See #1592 #1589
• [BUGFIX] Mesosphere: Cast error messages to strings. See #1614 [TEST?]
• [BUGFIX] PostgreSQL: Ignore rdsadmin database in PostgreSQL check to avoid permission error. See #1590 (Thanks @etrepum)
• [BUGFIX] PostgreSQL: Properly log bugs for custom metrics. See #1633 (Thanks @orenmazor)
• [BUGFIX] SQL Server: Fix collector freezes when connection is failing. See #1640
• [BUGFIX] SQL Server: Properly close cursor, avoid leaks. See #1631
• [BUGFIX] SQL Server: Send fractional metrics as floats. See #1616
• [BUGFIX] Varnish: Allow tags in varnish 3 XML style parsing. See #1645
• [BUGFIX] WMI: Cast tag values as strings. See #1630