Bot releases are visible (Hide)
Published by ogaca-dd about 4 years ago
Release on: 2020-07-22
6.21.0 ships the same features as 7.21.0 except for the Python versions it supports.
Please refer to the 7.21.0 changelog.
Release on: 2020-07-16
integration
integrations-extras
status
command now includes the flavor of the Agent thatautodiscovery_subnet
to available SNMP template extra configsconfig set dogstatsd_stats
command oflowercase_device_tag
option to the system io
core check ondevice
tag,system.io.*
metrics of Agent v5 andsystem.disk.*
metrics of all Agent versions.Fix missing values from cluster-agent status command.
Add missing device_name
tag in iostats_pdh
Fixes an issue where DD_TAGS were not applied to EKS Fargate pods
and containers.
Add freetds
linux dep needed for SQL Server to run in Docker
Agent.
APM : Fix parsing of non-ASCII numerals in the SQL obfuscator.
Previously unicode characters for which unicode.IsDigit returns true
could cause a hang in the SQL obfuscator
APM: correctly obfuscate AUTH command.
Dogstatsd standalone: when running on a systemd-based system, do not
stop Dogstatsd when journald is stopped or restarted.
Fix missing logs and metrics for docker-labels based autodiscovery
configs after container restart.
Fix bugs introduced in 7.20.0/6.20.0 in the Agent 5 configuration
import command: the command would not import some Agent config
settings, including api_key
, and would write some Docker &
Kubernetes config settings to wrongly-located files.
Fixes tag extraction from Kubernetes pod labels when using patterns
on certain non-alphanumeric label names (e.g.
app.kubernetes.io/managed-by).
Fixes the /ready health endpoint on the cluster-agent.
The /ready health endpoint was reporting 200 at the cluster-agent
startup and was then, permanently reporting 500 even though the
cluster-agent was experiencing no problem. In the body of the
response, we could see that a healthcheck component was failing.
This change fixes this issue.
This fix aims to cover the case when the agent is running inside GKE
with workload identity enabled. If workload identity is enabled,
access to /instance/name is forbidden, resulting into an empty host
alias.
Fix hostname resolution issue preventing the Process and APM agents
from picking up a valid hostname on some containerized environments
Fix a bug which causes certain configuration options to be ignored
by the process-agent
in the presence of a system-probe.yaml
.
Process agent and system probe now correctly accept multiple API
keys per endpoint.
The device_name
tag is not used anymore to populate the Device
field of a series. Only the device
tag is considered.
Fixes problem on Windows where ddagentuser home directory is left
behind.
Revert upgrade of GoSNMP and addition of extra authentication
protocols.
Add support for examining processes inside Docker containers running
under systemd cgroups. This also reduces agent logging volume as
it's able to capture those statistics going forward.
APM: The agent now exits with code 0 when the API key is not
specified. This is so to prevent the Windows SCM from restarting the
process.
1.13.11
.Published by prognant over 4 years ago
Release on: 2020-06-16
Published by truthbk over 4 years ago
Release on: 2020-06-11
Published by prognant over 4 years ago
6.20.0
ships the same features as 7.20.0
except for the Python versions it supports.
Please refer to the 7.20.0 changelog.
Published by prognant over 4 years ago
Release on: 2020-06-09
install_info
reportinstall_info
report and sends it through Host metadataservice
tag from Pod Metadata Labels.container_exclude_metrics
and container_include_metrics
can be used to filter metrics collection for autodiscovered containers. container_exclude_logs
and container_include_logs
can be used to filter logs collection for autodiscovered containers.device_name
tag on IOstats and disk checks.HOSTNAME_FQDN_ENABLED
to set the config value of hostname_fqdn
.device_name
tags on docker, containerd and network checks. Make series manage device_name
tag if device
is missing.ecr
, gcr
and mcr
are excluded automatically by the Agent./live
) fails in case of unrecoverable error that deserve an agent restart. (Ex.: goroutine deadlock or premature exit)/ready
) fails in case of recoverable errors or errors for which an agent restart would be more nasty than useful (Ex.: the forwarder fails to connect to DataDog)kube_cluster_name
and ecs_cluster_name
tags in addition to cluster_name
. Add the possibility to stop sending the cluster_name
tag using the parameter disable_cluster_name_tag_key
in Agent config. The Agent keeps sending kube_cluster_name
and ecs_cluster_name tags regardless of disable_cluster_name_tag_key.container_exclude
replaces ac_exclude
. container_include
replaces ac_include
. ac_exclude
and ac_include
will keep being supported but the Agent ignores them in favor of container_exclude
and container_include
if they're defined.system.swap.pct_free
metric.0.3.4
Published by KSerrania over 4 years ago
6.19.2
ships the same features as 7.19.2
except for the Python versions it supports.
Please refer to the 7.19.2 changelog.
Published by KSerrania over 4 years ago
Release on: 2020-05-12
Published by KSerrania over 4 years ago
Release on: 2020-05-04
This release is only an Agent v7 release, as Agent v6 is not affected by the undermentioned bug.
Published by albertvaka over 4 years ago
Release on: 2020-04-30
Default logs-agent to use HTTPS with compression when possible.
Starting from this version, the default transport is HTTPS with compression instead of TCP.
The usage of TCP is kept in the following cases:
To force the use of TCP or HTTPS with compression, logs_config.use_tcp or logs_config.use_http can be set to true, respectively.
The Agent is now available on Chocolatey for Windows
Add --full-sketches
option to agent check command that displays bins information
Support logs collection from Kubernetes log files with old Kubernetes versions (< v1.10).
Expose the new JMXFetch rate with metrics method to test JMX based checks.
The ac_include
and ac_exclude
auto-discovery parameters now support the
kube_namespace:
prefix to collect or discard logs and metrics for whole namespaces
in addition to the name:
and image:
prefixes to filter on container name and image name.
EKS Fargate containers now appear in the live containers view.
All processes running inside the EKS Fargate Pod appear in the live processes view
when shareProcessNamespace
is enabled in the Pod Spec.
Add the ability to change log_level at runtime. The agent command
has been extended to support new operation. For example to set
the log_level to debug the following command should be used:
agent config set log_level debug
, all runtime-configurable
settings can be listed using agent config list-runtime
. The
log_level may also be fetched using the agent config get log_level
command. Additional runtime-editable setting can easily be added
by following this implementation.
The system-probe
classifies UDP connections as incoming or outgoing.
Adds a new config option to the systemd core check. It adds the ability to provide a custom
mapping from a unit substate to the service check status.
The systemd core check now collects and submits the systemd version as check metadata.
Add host_provider_id
tag to Kubernetes events; for AWS instances this is unique in
contrast to the Kubernetes nodename currently provided with the host
tag.
On Windows, now reports system.io.r_await and system.io.w_await.
Metrics are reported from the performance monitor "Avg. Disk sec/Read" and
"Avg. Disk sec/Write" metrics.
Allow setting is_jmx
at the instance level, thereby enabling integrations
to utilize JMXFetch and Python/Go.
The authentication token file is now only created
when the agent is launched with the agent start
command
It prevents command such as agent status
to create
an authentication token file owned by a wrong user.
Count of successful DNS responses are tracked for each connection.
Network information is collected when the agent is running in docker (host mode only).
Make sure we don't recognize sha256:...
as a valid image name and add fallback to
.Config.Image in case it's impossible to map sha256:...
to a proper image name
Extract env, version and service tags from Docker containers
Extract env, version and service tags from ECS Fargate containers
Extract env, version and service tags from kubelet
Log configurations of type file
now accept a new parameter that allows
to specify whether a log shall be tailed from the beginning
or the end. It aims to allow whole log collection, including
events that may occur before the agent first starts. The
parameter is named start_position
and it can be set to
end
or beginning
, the default value is end
.
Resolve Docker image name using config.Image in the case of multiple image RepoTags
The agent configcheck command output now scrubs sensitive
data and prevents API keys, password, token, etc. to
appear in the console.
Errors that arise while loading checks configuration
files are now send with metadata along with checks
loading errors and running errors so they will show
up on the infrastructure list in the DataDog app.
Remove cgroup deps from Docker utils, allows to implement several backends for Docker utils (i.e. Windows)
On Windows, for Python3, add additional C-runtime DLLs to fix missing dependencies (notably for jpype).
Fix 'check' command segfault when running for more than 1 hour (which could
happen when using the '-b' option to set breakpoint).
Fix panic due to concurrent map access in Docker AD provider
Fix the flare
command not being able to be created for the non-core agents (trace,
network, ...) when running in a separated container, such as in Helm. A new
option, --local
, has been added to the flare
command to force the
creation of the archive using the local filesystem and not the one where
the core agent process is in.
Fix logs status page section showing port '0' being used when using the
default HTTPS URL. The status page now show 443.
Fix S6 behavior when the core agent dies.
When the core agent died in a multi-process agent container managed by S6,
the container stayed in an unhealthy half dead state.
S6 configuration has been modified so that it will now exit in case of
core agent death so that the whole container will exit and will be restarted.
On Windows, fixes Process agent memory leak when obtaining process arguments.
When a DNS name with ".local" is specifed in the parameter DDAGENTUSER_NAME, the correctly finds the corresponding domain.
Fix an issue where conf.yaml.example
can be missing from Add a check
menu in the Web user interface.
process-agent and system-probe now clean up their PID files when exiting.
When the HTTPS transport is used to send logs, send the sourcecategory as the sourcecategory:
tag
instead of ddsourcecategory:
, for consistency with other transports.
All Agents binaries are now compiled with Go 1.13.8
JMXFetch upgraded to 0.36.1. See 0.36.1 <https://github.com/DataDog/jmxfetch/releases/0.36.1>
_
and 0.36.0 <https://github.com/DataDog/jmxfetch/releases/0.36.0>
_
The statsd_metric_namespace
option now ignores the following metric
prefixes: airflow
, confluent
, hazelcast
, hive
, ignite
,
jboss
, sidekiq
Published by albertvaka over 4 years ago
Release on: 2020-04-30
Default logs-agent to use HTTPS with compression when possible.
Starting from this version, the default transport is HTTPS with compression instead of TCP.
The usage of TCP is kept in the following cases:
To force the use of TCP or HTTPS with compression, logs_config.use_tcp or logs_config.use_http can be set to true, respectively.
The Agent is now available on Chocolatey for Windows
Add --full-sketches
option to agent check command that displays bins information
Support logs collection from Kubernetes log files with old Kubernetes versions (< v1.10).
Expose the new JMXFetch rate with metrics method to test JMX based checks.
The ac_include
and ac_exclude
auto-discovery parameters now support the
kube_namespace:
prefix to collect or discard logs and metrics for whole namespaces
in addition to the name:
and image:
prefixes to filter on container name and image name.
EKS Fargate containers now appear in the live containers view.
All processes running inside the EKS Fargate Pod appear in the live processes view
when shareProcessNamespace
is enabled in the Pod Spec.
Add the ability to change log_level at runtime. The agent command
has been extended to support new operation. For example to set
the log_level to debug the following command should be used:
agent config set log_level debug
, all runtime-configurable
settings can be listed using agent config list-runtime
. The
log_level may also be fetched using the agent config get log_level
command. Additional runtime-editable setting can easily be added
by following this implementation.
The system-probe
classifies UDP connections as incoming or outgoing.
Adds a new config option to the systemd core check. It adds the ability to provide a custom
mapping from a unit substate to the service check status.
The systemd core check now collects and submits the systemd version as check metadata.
Add host_provider_id
tag to Kubernetes events; for AWS instances this is unique in
contrast to the Kubernetes nodename currently provided with the host
tag.
On Windows, now reports system.io.r_await and system.io.w_await.
Metrics are reported from the performance monitor "Avg. Disk sec/Read" and
"Avg. Disk sec/Write" metrics.
Allow setting is_jmx
at the instance level, thereby enabling integrations
to utilize JMXFetch and Python/Go.
The authentication token file is now only created
when the agent is launched with the agent start
command
It prevents command such as agent status
to create
an authentication token file owned by a wrong user.
Count of successful DNS responses are tracked for each connection.
Network information is collected when the agent is running in docker (host mode only).
Make sure we don't recognize sha256:...
as a valid image name and add fallback to
.Config.Image in case it's impossible to map sha256:...
to a proper image name
Extract env, version and service tags from Docker containers
Extract env, version and service tags from ECS Fargate containers
Extract env, version and service tags from kubelet
Log configurations of type file
now accept a new parameter that allows
to specify whether a log shall be tailed from the beginning
or the end. It aims to allow whole log collection, including
events that may occur before the agent first starts. The
parameter is named start_position
and it can be set to
end
or beginning
, the default value is end
.
Resolve Docker image name using config.Image in the case of multiple image RepoTags
The agent configcheck command output now scrubs sensitive
data and prevents API keys, password, token, etc. to
appear in the console.
Errors that arise while loading checks configuration
files are now send with metadata along with checks
loading errors and running errors so they will show
up on the infrastructure list in the DataDog app.
Remove cgroup deps from Docker utils, allows to implement several backends for Docker utils (i.e. Windows)
On Windows, for Python3, add additional C-runtime DLLs to fix missing dependencies (notably for jpype).
Fix 'check' command segfault when running for more than 1 hour (which could
happen when using the '-b' option to set breakpoint).
Fix panic due to concurrent map access in Docker AD provider
Fix the flare
command not being able to be created for the non-core agents (trace,
network, ...) when running in a separated container, such as in Helm. A new
option, --local
, has been added to the flare
command to force the
creation of the archive using the local filesystem and not the one where
the core agent process is in.
Fix logs status page section showing port '0' being used when using the
default HTTPS URL. The status page now show 443.
Fix S6 behavior when the core agent dies.
When the core agent died in a multi-process agent container managed by S6,
the container stayed in an unhealthy half dead state.
S6 configuration has been modified so that it will now exit in case of
core agent death so that the whole container will exit and will be restarted.
On Windows, fixes Process agent memory leak when obtaining process arguments.
When a DNS name with ".local" is specifed in the parameter DDAGENTUSER_NAME, the correctly finds the corresponding domain.
Fix an issue where conf.yaml.example
can be missing from Add a check
menu in the Web user interface.
process-agent and system-probe now clean up their PID files when exiting.
When the HTTPS transport is used to send logs, send the sourcecategory as the sourcecategory:
tag
instead of ddsourcecategory:
, for consistency with other transports.
All Agents binaries are now compiled with Go 1.13.8
JMXFetch upgraded to 0.36.1. See 0.36.1 <https://github.com/DataDog/jmxfetch/releases/0.36.1>
_
and 0.36.0 <https://github.com/DataDog/jmxfetch/releases/0.36.0>
_
The statsd_metric_namespace
option now ignores the following metric
prefixes: airflow
, confluent
, hazelcast
, hive
, ignite
,
jboss
, sidekiq
Published by hush-hush over 4 years ago
6.18.0
ships the same features as 7.18.0
except for the Python versions it supports.
Please refer to the 7.18.0 changelog.
Published by hush-hush over 4 years ago
Release on: 2020-03-13
service
defined in the init_config
orinstances
section of integration configuration, with the latterThe min_collection_interval check setting has been relocated since
Agent 6/7 release. The agent import command now include in the right
section this setting when importing configuration from Agent 5.
Add new config parameter (dogstatsd_entity_id_precedence) to
enable DD_ENTITY_ID presence check when enriching Dogstatsd
metrics with tags.
Add an option to exclude log files by name when wildcarding
directories. The option is named exclude_paths
, it can be added
for each custom log collection configuration file. The option
accepts a list of glob.
The status output now shows checks' last execution date and the last
successful one.
On debian- and rhel-based systems, system-probe is now set up so
that it can run in SELinux-enabled environments.
On Windows, a step to set the site
parameter has been added to the
graphical installer.
Added support for inspecting DNS traffic received over TCP to gather
DNS information.
Review the retry strategy used by the agent to connect to external
services like docker, kubernetes API server, kubelet, etc. In case
of failure to connect to them, the agent used to retry every 30
seconds 10 times and then, to give up. The agent will now retry
after 1 second. It will then double the period between two
consecutive retries each time, up to 5 minutes. So, After 10
minutes, the agent will keep on retrying every 5 minutes instead of
completely giving up after 5 minutes. This change will avoid to have
to restart the agent if it started in an environment that remains
degraded for a while (docker being down for several minutes for
example.)
Adds message field to the ComponentStatus check of
kube_apiserver_controlplane.up service check.
Add a config option ec2_use_windows_prefix_detection
to use the
EC2 instance id for Windows hosts on EC2.
Templates used for the agent status command are now embedded in the
binary at compilation time and thus original template files are not
required anymore at runtime.
Upgrade pip-tools
and wheel
dependencies for Python 2.
Improve interpolation performance during conversion from Prometheus
and OpenMetrics histograms to ddsketch
Allow sources for the Logs Agent to fallback to the service
defined in the init_config
section of integration configuration to
match the global tag that will be emitted.
Stop doing HTML escaping on agent status command output in order to
properly display all non-alphanumeric characters.
Upgrade embedded Python 3 to 3.8.1. Link to Python 3.8 changelog:
https://docs.python.org/3/whatsnew/3.8.html
Note that the Python 2 version shipped in Agent v6 continues to be
version 2.7.17 (unchanged).
Removing an RPM of the Datadog Agent will no longer throw missing
files warning.
The agent config command output now scrubs sensitive data and
prevents API keys, password, token, etc. from appearing in the
console.
Add support for the EC2 instance metadata service (IMDS) v2 that
requires to get a token before any metadata query. The agent will
still issue unauthenticated request first (IMDS v1) before switching
to token-based authentication (IMDS v2) if it fails.
system-probe no longer needs to be enabled/started separately
through systemctl
systemd
core check handle gracefully missing SystemState
1.1.1d
, pyopenssl 19.0.0
and9.4.25
.>= 6.18.0
running Python 3 and version >= 7.18.0
are nowPublished by albertvaka over 4 years ago
Release on: 2020-02-26
This is a Windows-only release.
Published by julien-lebot over 4 years ago
6.17.1
ships the same features as 7.17.1
except for the Python versions it supports.
Please refer to the 7.17.1 changelog.
Published by julien-lebot over 4 years ago
Release on: 2020-02-20
Published by julien-lebot over 4 years ago
6.17.0
ships the same features as 7.17.0
except for the Python versions it supports.
Please refer to the 7.17.0 changelog.
Published by julien-lebot over 4 years ago
Release on: 2020-02-04
Change agents base images to Debian bullseye
Starting with this version, the containerized Agent never chooses
the OS hostname as its hostname when it is running in a dedicated
UTS namespace. This is done in order to avoid picking container IDs
or kubernetes POD names as hostnames, since these identifiers do not
reflect the identity of the host they run on.
This change only affects you if your agent is currently using a
container ID or a kubernetes POD name as hostname. The hostname of
the agent can be checked with agent hostname
. If the output stays
stable when the container or POD of the agent is destroyed and
recreated, you’re not impacted by this change. If the output
changes, it means that the agent was unable to talk to EC2/GKE
metadata server, it was unable to get the k8s node name from the
kubelet, it was unable to get the hostname from the docker daemon
and it is running in its dedicated UTS namespace. Under those
conditions, you should set explicitly define the host name to be
used by the agent in its configuration file.
kubernetes.io/role
label.statsd_metric_namespace
to dogstatsddatadog.tracer
.log_to_console
, log_to_syslog
and all other syslogasm goto
on Ubuntu 19.04outgoing
in containerizedcurl
to 7.66.0
, autoconf
to2.69
, procps
to 3.3.16