5.11.1 / 02-13-2017

5.11.1 / 02-13-2017

Linux and Docker only

Details

There was a regression in 5.11.0 due to an inadverted subdependency upgrade, breaking the SNMP check.

Changes

• [REGRESSION] pinning of pyasn1 to version 0.1.9. See #3183, omnibus-software-111, omnibus-software-114, omnibus-software-115.

5.11.0 / 02-08-2017

5.11.0 / 02-08-2017

Linux and Docker only

Details

This release bundles the datadog-trace-agent as part of Linux and Docker installs. The dd-agent codebase contains a minimal set of changes to ensure the new trace-agent process will plug neatly into existing agent installs. Beta customers of Datadog's tracing tool will have received an email detailing the steps to transition to dd-agent 5.11.0 from the deprecated dd-trace-agent package.

For more information:
https://github.com/DataDog/datadog-trace-agent
https://github.com/DataDog/datadog-trace-agent/releases/tag/5.11.0
5.10.1...5.11.0

Changes

• [IMPROVEMENT] Supervisor: Add trace-agent as an optional process managed by supervisor
• [IMPROVEMENT] Flare: Add trace-agent log file to flare output
• [IMPROVEMENT] Core: append the output of trace-agent -info to service datadog-agent info command

5.10.1

5.10.1 / 11-21-2016

Linux, Windows, Docker and Source Install

Details

5.10.0...5.10.1

Updated Integrations

• RiakCS
• Mongo

Changes

• [IMPROVEMENT] Core/Forwarder: stop flushing after 10s. See #3018.
• [IMPROVEMENT] Core: isolate system checks. See #3011.
• [IMPROVEMENT] RiakCS: support Riak CS 2.1+ stats format. See #2920. (Thanks @millerdev)
• [IMPROVEMENT] Status: Silence requests exception. See #3023.
• [IMPROVEMENT] SpooledTemporaryFile for subprocess output. See #3002.
• [BUGFIX] Core: fix unintended subprocess_output empty output errors. See #3024.
• [BUGFIX] Core/Multiple Checks: Only set psutil.PROCFS_PATH once in the collector. See #3013.
• [BUGFIX] Core: use proxy for API key status check in info page. See #3012. (Thanks @2rs2ts)
• [BUGFIX] Mongo: use db.current_op instead of manually querying. See #3016 (Thanks @ebroder)
• [BUGFIX] Mongo: use currentOp for monodb 3.1+. See #3015 (Thanks @lattwood)
• [DEPRECATE] Process: procfs_path is now deprecated, should be set in datadog.conf. See #3013.

5.10.0

5.10.0 / 11-09-2016

Linux, Windows, Docker and Source Install

Details

5.9.1...5.10.0

Kubernetes: metrics from the Kube State API

The new kubernetes_state check pulls metrics from the Kubernetes State API.

See #2890, #2984 and omnibus-software-85 for more details.

RPM package for SUSE

The Agent is now available as an RPM package for SUSE. Please follow the instructions on our
Agent installation page to use it.

For more details, see related PRs #2900, #2937, omnibus-software-72, dd-agent-omnibus-90

Linux packages: change of ownership of /opt/datadog-agent directory

The Linux packages (deb and rpm) now make the dd-agent user and group own the /opt/datadog-agent folder and
its contents when they're installed. This shouldn't cause any change on regular installations of the Agent,
but please take this change into account if you customize the ownership/permissions of the /opt/datadog-agent directory or
its contents.

For more details, see related PR: dd-agent-omnibus-97

Python dependencies upgrade

We have updated the following python dependencies shipped with the Agent. If you happen to run any
custom checks that rely on the former versions please make necessary amends:

• kazoo 2.2.1 (previously 1.3.1). See Kazoo's changelog.
• psutil 4.4.1 (previously 3.3.0). See psutil's changelog.
• curl 7.51.0 (previously 7.50.0). See curl's changelog.

Updated integrations

• Apache
• Ceph
• Consul
• Docker
• Elasticsearch
• FluentD
• Haproxy
• Http check
• Kafka
• Kubernetes
• Marathon
• Mesos
• Mongo
• MySQL
• Openstack
• PGBouncer
• Postgres
• Process
• RabbitMQ
• Redis
• Spark
• SSH check

Changes

• [FEATURE] Core: Add support for sum histogram_aggregates type. See #2780 (Thanks @ross)
• [FEATURE] Core: Display api key status on the info page. See #2947
• [FEATURE] Consul: Add net latency metrics. See #2757 (Thanks @ross)
• [FEATURE] Haproxy: Allow per-instance custom tags. See #2649, #2951 (Thanks @theojulienne)
• [FEATURE] Kubernetes: Add CPU and memory capacity reporting. See #2935 (Thanks @markine)
• [FEATURE] Mongo: Add metrics for fsyncLocked-ness, replset votes and vote fraction. See #2975 (Thanks @ebroder)
• [FEATURE] Process: Support pid file. See #2607 (Thanks @n0ts)
• [FEATURE] Spark: Support clusters provisioned using Apache Mesos. See #2797 (Thanks @zachradtka)
• [IMPROVEMENT] Ceph: Handle HEALTH_WARNING status. See #2929 (Thanks @ccocsas)
• [IMPROVEMENT] Consul: Remove warning when no whitelist is defined. See #2955
• [IMPROVEMENT] Consul: Support multiple instances. See #2948
• [IMPROVEMENT] Core: forwarder: Do not requeue all responses. See #2840
• [IMPROVEMENT] Core: collector: Catch exceptions raised by system checks. See #2973
• [IMPROVEMENT] Docker: Set alert_type to error for oom and kill events. See #2866
• [IMPROVEMENT] Docker: Set exec events' priority to low. See #2960
• [IMPROVEMENT] Docker/Kubernetes: Filter containers consistently. See #2836
• [IMPROVEMENT] Elasticsearch: Add support for ES 5.x. See #2989 (Thanks @ptinsley and @alexef)
• [IMPROVEMENT] FluentD: Add collection timeout. See #2978 (Thanks @mattrobenolt)
• [IMPROVEMENT] Kafka: Document specifying multiple Kafka / ZK nodes. See #2956 (Thanks @jeffwidman)
• [IMPROVEMENT] Kafka: Upgrade kazoo dep to 2.2.1. See #2964, omnibus-software-84
• [IMPROVEMENT] Kubernetes: Use the env var KUBERNETES_SERVICE_HOST if present. See #2921 (Thanks @arohner)
• [IMPROVEMENT] Kubernetes: Add namespace tag to kubernetes.pods.running. See #2931
• [IMPROVEMENT] Kubernetes: Add some tags to events. See #2926
• [IMPROVEMENT] Marathon: Count running/pending deployments. See #2788 (Thanks @bradhe)
• [IMPROVEMENT] Packaging: Upgrade psutil to 4.4.1. See #2957, omnibus-software-82
• [IMPROVEMENT] Packaging: Upgrade version of datadogpy shipped with Linux packages. See omnibus-software-80
• [IMPROVEMENT] PGBouncer: Support SSL connections to PGBouncer. See #2885, #2967 (Thanks @dblackdblack)
• [IMPROVEMENT] Postgres: Allow UNIX socket path as host config value. See #2734, #2968 (Thanks @DylanFrese)
• [IMPROVEMENT] RabbitMQ: Add option to skip SSL verification. See #2991 (Thanks @iancward)
• [IMPROVEMENT] Redis: Fix tags on can_connect service check. See #2803 (Thanks @gphat)
• [IMPROVEMENT] Redis: Enable monitoring of SSL-ified redis servers. See #2674 (Thanks @dblackdblack)
• [IMPROVEMENT] SSH Check: Support ECDSA keys (<= 256bits). See #2965 (Thanks @n0ts)
• [BUGFIX] Apache: Add timeout on connection to Apache status URL See #2928
• [BUGFIX] Core: Avoid CPU spikes caused by subprocess calls. See #2950
• [BUGFIX] Core: Catch EOFError when loading info page. See #2972
• [BUGFIX] Core: Display expected service check count. See #2938
• [BUGFIX] Core: Handle timezone encodings on Windows. See #2942
• [BUGFIX] Core: Ignore all non-unicode-decodable characters in collector's emitter. See #2941
• [BUGFIX] Core: In dev mode, fallback to /tmp to dump pstats. See #2934
• [BUGFIX] Core: On the Windows UI, set color to white for background of info page. See #2970
• [BUGFIX] Core: On Windows, retry on process restart failures. See #2897
• [BUGFIX] Kubernetes: Fix kubelet health check request. See #2887 (Thanks @yannrouillard)
• [BUGFIX] Kubernetes: Fix parsing of units in Kubernetes resource quantities. See #2908 (Thanks @therc)
• [BUGFIX] Http Check: Fix SSL certificate parsing issue related to strptime. See #2915 (Thanks @ghTravis)
• [BUGFIX] Mesos Slave: Connect to master with same protocol as slave. See #2910 (Thanks @mcclurmc)
• [BUGFIX] Service Discovery: Fix single check reload on auto_conf templates and removed containers. See #3006
• [BUGFIX] Mongo: Fix case when check would send password in clear. See #2940
• [BUGFIX] Mongo: Fix hostname detection when username/password are part of the uri. See [#2894](#2...

5.9.1

5.9.1 / 09-30-2016

Linux, Docker and Source Install

Details

5.9.0...5.9.1

Build/Packaging Issue in 5.9.0

A packaging issue led to issues with the Spark integration when running Agent 5.9.0. This release fixes the issue.

Updated integrations

• Kubernetes
• Spark

Changes

• [BUGFIX] Dogstatsd: fix ipv6 fallback to ipv4. See #2879
• [BUGFIX] Kubernetes: Fix incorrect time conversion for Kubernetes events. See #2873 (Thanks @yannrouillard)
• [BUGFIX] Packaging: Build with the corresponding dd-agent-omnibus branch ( 5.9.x )
• [BUGFIX] Packaging: include latest repo keys. See above.
• [BUGFIX] Spark: fix missing libraries issue.

5.9.0

5.9.0 / 09-27-2016

Linux, Windows, Docker and Source Install

Details

5.8.5...5.9.0

Kafka-Python library update

Because we have bumped the library version for the kafka-python package, if you happen to run any custom
checks that rely on the former version please make necessary amends. We apologize for any inconvenience.
See #2709 for reference.

Updated integrations

• Apache
• Cassandra
• Ceph
• Disk Check
• DNS
• Docker
• Elasticsearch
• Gearman
• HTTP Check
• IIS
• JMXFetch
• Kafka Consumer
• Kubernetes
• Linux Check
• Marathon
• Mesos
• Mongo
• MySQL
• Network Check
• OpenStack
• Php5_fpm
• Postgres
• Process
• RabbitMQ
• Redis
• Spark
• SSH
• vSphere
• Windows

Deprecated integrations

• Jenkins

Changes

• [ENHANCEMENT] Use JMXFetch 0.12.0 ( Changelog )
• [SECURITY] The deb and rpm packages now bundle OpenSSL 1.0.2i. For more details, see the security advisory, omnibus-software-76
• [FEATURE] Apache: add connection status metrics. See #2622
• [FEATURE] Core: dogstatsd: Added support for IP agnostic connections. See #2619
• [FEATURE] Curl: support for HTTP2. See omnibus-software-65
• [FEATURE] Dns_check: Allow querying a custom record type. See #2653
• [FEATURE] Docker_daemon: resolve hierarchy by looking at procfs /cgroup. See #2537
• [FEATURE] Forwarder: allow multiple endpoints/api_keys. See #2564
• [FEATURE] Gearman: Add per-task gearman metrics. See #2672 (Thanks @nyanshak)
• [FEATURE] Http_check: Add allow_redirects parameter. See #2586 (Thanks @n0ts)
• [FEATURE] Http_check: Add ability to use post method. See #2544
• [FEATURE] Http_check: SNI support for cert_expiration. See #2521
• [FEATURE] Kubernetes: Improve service discovery to only reload checks that need it. See #2551
• [FEATURE] Kubernetes: Ingest k8s events + limits and requests metrics. See #2551
• [FEATURE] Linux: make /proc configurable. See #2482
• [FEATURE] Mongo: Apply yaml-configured tags to service checks. See #2575 (Thanks @avaughan89)
• [FEATURE] Mysql: Allow connection_timeout to be set for MySQL checks. See #2729 (Thanks @scottgeary)
• [FEATURE] Process: Option to search processes by PID. See #2119 (Thanks @handigarde)
• [FEATURE] Spark: spark standalone support. See #2752 (Thanks @zachradtka), omnibus-software-73, dd-agent-omnibus
• [IMPROVEMENT] Cassandra: add read and write latency ColumnFamily metrics. See #2483
• [IMPROVEMENT] Ceph: improve health metrics. See #2852
• [IMPROVEMENT] Ceph: update to support 10.0.2.2. See #2805
• [IMPROVEMENT] Core: avoid calls to service discovery from dogstatsd. See #2798
• [IMPROVEMENT] Core: easier config for multiple endpoints. See #2774
• [IMPROVEMENT] Core: Add an HTTP timeout to many integrations. See #2673 (Thanks @gphat)
• [IMPROVEMENT] Core: expiration should adjust to longer collection intervals (>300s). See #2541
• [IMPROVEMENT] Core: remove events & resources api_key. See #2557
• [IMPROVEMENT] Core: remove noisy logs. See #2715 (Thanks @ensonik)
• [IMPROVEMENT] Disk: handle multilines df output. See #2733
• [IMPROVEMENT] Docker: Improve service discovery to only reload checks that need it. See #2702
• [IMPROVEMENT] Docker: Add CPU throttling stats to Docker metrics. See #2724 (Thanks @ejholmes)
• [IMPROVEMENT] Dogstatsd: fix server address when non_local_traffic is passed. See #2691
• [IMPROVEMENT] Elasticsearch: Add missing metrics. See #2758 (Thanks @mdelaney)
• [IMPROVEMENT] Elasticsearch: Add elastic search cluster name to tags. See #2744, #2820 and #2826
• [IMPROVEMENT] Flare: Close the tar file cleanly once upload is done. See #2621
• [IMPROVEMENT] Flare: hide multiple endpoints api_keys. See #2646
• [IMPROVEMENT] Gohai: Processes - simplify payload. See #2600
• [IMPROVEMENT] Gohai: CPU - take into account environments with multiple physical processors. See gohai-32
• [IMPROVEMENT] Http_check: log exceptions 🔊. See #2704
• [IMPROVEMENT] Kafka_consumer: bumping kafka-python package version to 1.2.5. See #2709, omnibus-software-70
• [IMPROVEMENT] Kubernetes: Add ability to specify service discovery via kubernetes annotations. See #2848 (Thanks @mikekap)
• [IMPROVEMENT] Kubernetes: disable use_histogram. See #2542
• [IMPROVEMENT] Kubernetes: Log URL in kubelet check failures and service checks. See #2735 (Thanks @therc)
• [IMPROVEMENT] Mesos: make SSL check optional. See #2809
• [IMPROVEMENT] Mongo: Adds metrics for collections. See #2739
• [IMPROVEMENT] Multiple integrations: Add an HTTP timeout to many integrations. See #2673 (Thanks @gphat)
• [IMPROVEMENT] Network: implement check using psutil on Windows. See #2499
• [IMPROVEMENT] Openstack: openstack should pick-up on proxy settings set on agent. See #2572
• [IMPROVEMENT] Openstack: nova/neutron extend name search. See #2627
• [IMPROVEMENT] Packaging: debian - only create dd-agent group if it doesn't already exist. See dd-agent-omnibus-89
• [IMPROVEMENT] Packaging: linux - correctly silence configcheck. See dd-agent-omnibus-86
• [IMPROVEMENT] Packaging: debian - remove configcheck output when upgrading. See dd-agent-omnibus-85
• [IMPROVEMENT] Php5_fpm: Allow host header override. See #2779 (Thenks @allixsenos)
• [IMPROVEMENT] Postgres: Allow to use psycopg2. See #2782
• [IMPROVEMENT] Process: Add metrics for avg, max, and min run times for processes. See #2531
• [IMPROVEMENT] Process: Add system.processes.mem.pct metric to process check. See #2801
• [IMPROVEMENT] Python: Upgrade to 2.7.12. See omnibus-software-63
• [IMPROVEMENT] RabbitMQ: Avoiding the extra double lookup for 'rabbitmq_api_url'. See #2543 (Thanks @ParthKolekar)
• [IMPROVEMENT] RabbitMQ: new health check. See #2711
• [IMPROVEMENT] SSH: Be more specific when logging ssh errors. See #2708
• [IMPROVEMENT] Util: remove LaconicFilter. See #2605
• [IMPROVEMENT] Vsphere: Add optional vm include parameter. See #2459
• [BUGFIX] Core: forwarder drop payloads when no valid endpoints. See #2833
• [BUGFIX] Core: emitter would crash agent with some unicode control characters in payload. See #2843
• [BUGFIX] Core: service discovery fix fallback IP address extraction. See #2855
• [BUGFIX] Core: service discovery dont pass missing config_store reference to dockerutil. See #2858
• [BUGFIX] Cloud metadata: Fail on non-2xx responses. See #2844
• [BUGFIX] Disk: timeout on disk usage. See [#2714][] and #2823
• [BUGFIX] Docker_daemon: ECS introspection resilience. See #2745 and #2825
• [BUGFIX] Docker_daemon: Custom cgroups are per instance only. See #2846
• [BUGFIX] Docker_daemon: Fix pid retrieval that could affect k8s. See #2847
• [BUGFIX] Dockerutil: cpu subsys might be mistaken for cpuacct. See #2829
• [BUGFIX] Http_check: Bring back include_content option. See #2631
• [BUGFIX] IIS: Fix metrics tagging when multiple sites are specified on instance. See #2677
• [BUGFIX] JMXFetch: Do not scope MBeans queries on list_not_matching_attributes action. See jmxfetch-102
• [BUGFIX] Marathon: Fix a small problem that prevented marathon full path from being properly built. See #2620
• [BUGFIX] Mongo: Repairing of mongodb.can_connect check. See #2658 (Thanks @cryptspirit)
• [BUGFIX] Mongo: Fixes a mistake in the mongo collections check. See #2783
• [BUGFIX] MySQL: Check for null key_buffer_size before dividing by it. See #2784
• [BUGFIX] Network Checks: Network check instance corrections. See #2736 (Thanks @tschellhorn)
• [BUGFIX] Packaging [linux]: Kernel 4+ won't error out on status now. See #2640
• [BUGFIX] Packaging [osx]: Support GNU-Friendly sed. See #2580 (Thanks @sha1sum)
• [BUGFIX] Process: Use configured proc path when getting pagefault stats #2727 (Thanks @rdxiang)
• [BUGFIX] Redisdb: Avoid exception when slowlog query returns empty command. See #2651
• [BUGFIX] Win32_event_log: Fix small regression on msg_text selection. See #2591
• [BUGFIX] Win32_event_log: reset filters between iterations. See #2796
• [BUGFIX] Wmi: set provider architecture when necessary. See #2812
• [DEPRECATED] Core: clean debug logs and kill Dogstream-based DdForwarder. See #2628
• [DEPRECATED] Jenkins: Deprecate check. See #2688
• [DEPRECATED] Nghttp2: remove python bindings. See omnibus-software-71

5.8.5

5.8.5 / 07-27-2016

Windows, Linux and Source Install

Details

5.8.4...5.8.5

Future rotation of the APT signing key

In preparation for a future rotation of our package signing keys, the 5.8.5 DEB package will, on install, import a new
trusted APT key:

pub   4096R/382E94DE 2016-06-29 [expires: 2022-06-28]
uid                  Datadog, Inc <[email protected]>

During the package install, the DEB package will output the following:

Prepare Datadog Agent keys rotation
  Add the new 'Datadog, Inc <[email protected]>' key to the list of APT trusted keys. ... OK

The signing key of the Agent hasn't changed yet but will be switched to this new key in a future release.

See dd-agent-omnibus-81

Changes

• [IMPROVEMENT] Core: Upgrade embedded python to 2.7.12. See omnibus-software-63
• [IMPROVEMENT] Elasticsearch: Take into account node name for cluster stats. See #2696
• [IMPROVEMENT] RPM package: Add runtime dependency on the initscripts package. See dd-agent-omnibus-82
• [IMPROVEMENT] RPM package: Don't delete dd-agent user and group on uninstall. See dd-agent-omnibus-84
• [BUGFIX] Core: Use flag to signal config reload to avoid race conditions. See #2670
• [BUGFIX] Core: Don't use Docker hostname if it's an EC2 one. See #2661
• [BUGFIX] Docker: Fix disk metrics rounding issue. See #2626
• [BUGFIX] Haproxy: Fix KeyError when an unknown status is found. See #2681
• [BUGFIX] IIS: Remove warnings on 'Name' property. See #2633
• [BUGFIX] MongoDB: Fix case where optimeDate is not available. See #2625
• [BUGFIX] Service Discovery: Introduce _get_image_ident and fix a bug that needed it. See #2684
• [BUGFIX] Windows Event log: Support unicode messages. See #2660

5.8.4

5.8.4 / 07-08-2016

Windows, Linux and Source Install

Details

5.8.3...5.8.4

Changes

• [IMPROVEMENT] Core: Upgrades supervisord to 3.3.0 See [#2667][], [#2669][]
• [BUGFIX] MySQL: Fixes MySQL slave detection. See [#2610][]
• [BUGFIX] MySQL: Fixes MySQL replication service check. See [#2603][]
• [BUGFIX] Core: Fixes a bug that caused the thread pool to leak threads. See [#2666][]

5.8.4-rc.1

[tests] Force use of fluent 0.12.22

0.12.24 has some dependency issues

5.8.3

5.8.3 / 07-05-2016

Windows, Linux and Source Install

Details

5.8.2...5.8.3

Changes

• [FEATURE] Flare: Adds configcheck output to flare command. See [#2588][]
• [IMPROVEMENT] Docker: Bump docker-py to 1.8.1 for network features support. See [#2556][]
• [BUFIX] HAProxy: Add collate_status_tags_per_host flag See [#2590][]
• [BUGFIX] Core: A bug prevented ipv6 from being used in some circumstances. This fixes that, and defaults to ipv6 and falls back to ipv4. See [#2592][]
• [BUGFIX] Docker: Handles buggy responses from docker api better. See [#2608][]
• [BUGFIX] MongoDB: Only collect ReplicationInfo when it's possible to do so, rather than erroring out. See [#2559][]
• [BUGFIX] Postgres: Adds a hard limit on postgres.table.count metric, as this can be very expensive. See [#2575][]
• [BUGFIX] PowerDNS Recursor: The configuration file needed to be renamed from powerdns.conf to powerdns_recursor.conf. See [#2538][]
• [BUGFIX] Service Discovery: Improvements for testing, logging and service variable interpolation. See [#2573][]
• [BUGFIX] Service Discovery: Yse docker hostname rather than default route to query cadvisor and kublet. See [#2609][]
• [BUGFIX] Service Discovery: Use get_identifier instead of buggy image name extraction. See [#2593][]
• [BUGFIX] SQLServer: Send service checks after every run, rather than only at the beginning. See [#2515][]
• [BUGFIX] vSphere: Enhances topology support, skip unknown metrics. See [#2560][]
• [BUGFIX] vSphere: The whole check shouldn't fail just because the check failed on a certain instance. See [#2548][]
• [BUGFIX] Win32: When memory check collection times out, it causes an error in the collector. Instead, it should recover from this. See [#2553][]
• [BUGFIX] WMI: Allows user to set a profider in request data. See [#2565][], [#2369][]