Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- a7b17c6 Inventory data collection could now work in a non-blocking manner: We now allocate a configurable inventory queue in the same fashion as metrics/events. This feature is disabled under a feature-flag.
Fixed
- 71cc96b Fixed issue in the storage sampler where we reported only one mount point for each device.
Changed
- 14bcf94 The built-in Docker integration has been updated to version 1.3.2. For more information, see the nri-docker changelog
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Fixed
- PR-50 Fixed inventory deltas submission for remote (non host) entities. This fixes inventory errors showing up on the log for
DEBUG
level with messageerror field deltas not in expected order: have delta ID XXX, want YYY
.
Context
Inventory deltas are JSON diffs and they need to be applied in order aiming to provide a consistent state. Both the infrastructure agent and the New Relic database stored their latest delta ID received/submitted.
This delta ID is incremented in each plugin entity inventory submission to New Relic. When these IDs go out of sync between agent and platform, the latter provides its latest known ID for the given inventory plugin and entity. Then a reconciliation process is run by the agent for that plugin so that the next ID will match the one expected by New Relic.
Root cause
For remote, non-host entities the agent side didn't store delta IDs on a per-entity basis as the New Relic database did. This means that for a given inventory plugin or integration providing more than 1 entity there'd be delta ID order issues, which were only seen on agent's DEBUG level (verbose mode > 0). At some point all data would usually be submitted as reconciliation process attempts to fix this. But for high load integrations producing multiple remote entities this reconciliation process would be triggered every time an entity needs to be submitted as for each of them platform keeps a record, while the agent recorded ID on a per-plugin basis. Therefore, for high-load integrations, the agent reconciliation process might enter in race-condition loop when trying to fix the delta IDs.
Fix
This release fixes this behaviour by keeping track of delta IDs in the same way as the New Relic database does (or a per-plugin, per-entity basis).
As state might be stored on disk for inventory deltas, if any integration had it persisted before the fix it'll take a reconciliation run per entity to fix agent's locally stored IDs. In this case you might still see some log entries, but they'll be gone during their first submissions.
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- Beta version (v0.1.0-beta) of nri-winservices is now packaged with the agent. For more information, see the Windows services integration documentation.
Fixed
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Processes metrics can be enabled for new accounts
By default, the Infrastructure agent doesn't send data about the operating system's processes. The agent still collects such data, unless metrics_process_sample_rate
is set to -1
.
The sending of processes data to New Relic can be manually enabled by setting enable_process_metrics
to true
. To fine-tune which processes you want to monitor, configure include_matching_metrics
or exclude_matching_metrics
.
Accounts and subaccounts created before July 20, 2020 have
enable_process_metrics
set totrue
by default; this can be overridden by setting a different value in the config file.
Added
- f670501 Added
ReadWriteBytesPerSecond
to storage sampler.
Fixed
- 1c06e93 Improved the handling of child process on Windows.
- 3394abd Fixed a bug where Logs in Context wouldn't map to an application when using the Infrastructure forwarder
- Updated KMS Secrets management to work correctly with
credential_file
andconfig_file
. - Removed some potential race conditions in the agent which should improve the stability of the agent.
- Fixed issue where storage sample would sometimes not report storage devices correctly.
Changed
- 04335dd Updated base Docker image for
newrelic/infrastructure
(used bynewrelic/infrastructure-k8s
) to Alpine3.12
and also updated the packages to latest versions in the container. - The built-in Flex integration has been updated to version 1.3.2. For more information, see the Flex changelog.
- The built-in Docker integration has been updated to version 1.3.1. For more information, see the nri-docker changelog
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
SystemSample
has two new attributes,MemoryUsedPercent
andMemoryFreePercent
.- The log forwarder now supports the
pattern
parameter fortcp
andsyslog
For more information on the log forwarder feature, see Forward your logs using New Relic Infrastructure.
Changed
- The built-in Flex integration has been updated to version 1.3.1. For more information, see the Flex changelog.
- Fluent Bit Output plugin for New Relic has been upgraded to v1.3.0. For more information, see the changelog.
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- The agent now adds Alibaba Cloud metadata, when available, to all data collected.
- When the agent runs in verbose mode, all integrations that support it now also run in verbose mode.
- When querying EC2 Instance Metadata Service the agent now uses Instance Metadata Service v2 (IMDSv2) for all requests. For more information on why you should use IMDSv2, see this AWS blog post .
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- The agent now logs
error
andfatal
messages from integrations even when not running in verbose mode. This feature is only available to integrations configured using the V4 configuration format found in versions 1.8.0 or higher of the agent.
Fixed
- Fixed an issue where spaces in integration commands would sometimes get split incorrectly, causing the commands to fail.
Changed
- The built-in Flex integration has been updated to version 1.3.0 For more information, see the Flex changelog
- The built-in Docker integration has been updated to version 1.3.0. For more information, see the nri-docker changelog
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Security fixes
- In Linux, at startup, the agent ensures it has ownership of the integrations temporary data folder, /tmp/nr-integrations/. This ensures only the agent/integrations can write into it negating potential abuse that could result in Denial/Degradation of Service.
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- The Log Forwarder now adds additional validation and information logging when parsing the logging configuration files.
- Example configuration for the Log Forwarder has been added to the
logging.d
directory in the Infrastructure agent. - When the Log Forwarder is started it will now only read YAML configuration files (*.yml and *.yaml).
For more information on the Log Forwarder, see Forward your logs using New Relic Infrastructure.
Fixed
- The Log Forwarder can now be started even if the hostname cannot be resolved.
Changed
- The built-in Flex integration has been updated to version 1.2.2. For more information, see the Flex changelog.
- Fluent Bit Output for New Relic has been increased to 1.2.1, see the changelog
Notes
A new version of the agent has been released. Follow standard procedures to update your Infrastructure agent.
Added
- HTTP proxy support for the log forwarder. Log records are automatically sent through the same proxy used by the agent to send metrics.
- [BETA] HTTPS proxy support for the log forwarder.