- The agent now reacts to failed New Relic requests based on response code, as opposed to parsing an exception message in the response body.
- Fixed clearing of active harvest via _stopHarvester()
- Fixed handling of harvest endpoints when not all fail.
- Added agent state "connecting" to indicate when handshake with New Relic servers is starting. This can be triggered on startup and restarts.
npm auditin security checks.
- Collector now specify
application/jsoncontent-type when data is compressed instead of
- Bumped ecmaVersion in test .eslintrc to 8
Added IP address collection and forwarding of metadata headers for upcoming protocol 17.
These features are currently behind the
protocol_17feature flag until all parts of protocol 17 are implemented.
Converted error handling in
Previously many of the errors were thrown. For consistency with async errors, these are now handed to the callback instead of thrown. The old behavior could result in a crash under a few circumstances, such as when the agent exceeded a configured maximum payload size. These errors came from
RemoteMethod._safeRequest. Since these errors are handed to the callback instead of thrown, this bug is no longer a potential.
Refactored harvest interactions in preparation for protocol 17 status codes.
Flag will be removed and protocol will be hard-coded to
17once functionality is released on New Relic backend.
This release also includes changes to the agent to enable monitoring of Lambda functions. If you are interested in learning more or previewing New Relic Lambda monitoring please email email@example.com.
This new option will make the agent wait for all active transactions to finish before actually shutting down. This does not pre-empt creation of new transactions, so care must be taken to ensure the active transaction pool drains or the agent will never shut down.
- Added switch statement indenting standard to eslintrc
- Introduced "warn" level 2 space rule to eslintrc
Changed totalTime attribute to be in decimal seconds instead of milliseconds for transaction events.
Agent no longer produces spans on ignored transactions.
Previously, the agent would produce distributed tracing span events regardless of the ignored status of the transaction the events originated from.
Extended Restify instrumentation to mark possible transaction names in order to account for async response methods.
hapi@16versioned tests to only run on Node 6 and above.
Pinned mysql2 to
<1.6.2in versioned tests.
This new method can be used to extract the database name from
Added link to CONTRIBUTING.md file in README.md
Thanks to Yuri Tkachenko (@tamtamchik) for the contribution.
Added VS Code settings to git ignore.
Fixed bug preventing Distributed Tracing (DT) from fully functioning when Cross Application Tracing (CAT) was disabled.
Renames better-cat integration test organization to be distributed-tracing and updated some test verbiage to use DT or distributed tracing instead of CAT or cross application tracing.
The agent will no longer break express routers in the case of using regex paths in a list.
Previously, the agent would overwrite the regex with the source of the regex. The agent now makes a copy of the route array and mutates that instead.
Attributes will now be properly propagated to PageView events.
The agent may now be configured to pass attributes along to the browser agent. The attributes that match the include/exclude rules in the
browser_monitor.attributessection will now be placed on PageView events.
superagentas built-in instrumentation.
This instrumentation just maintains transaction state when using the
superagentmodule to make HTTP requests with either callbacks or promises.
noticeErrorAPI method to be partially functional in High Security Mode.
In HSM, any custom attributes will be ignored, but the error will still be tracked. This brings the Node agent in line with the behavior of other language agents.
PromiseShimclass for instrumenting promise libraries.
Support for setting
transaction_tracer.transaction_thresholdto 0 has been added.
- Updated DT payload creation to use
primary_application_idfrom connect response.
- Added protection against functions with modified prototypes in
- Replaced SQL ID hash generation algorithm with SHA1 instead of MD5 to allow usage in FIPS compliant environments.
- Leveraged 16 hex digits for creation of SQL ID.
- Upgraded ejs module to get rid of Github security warnings. The ejs module was only used for tests and not in main agent code.
- Changes DT payload configuration log messages to debug level as it is not uncommon for calls to occur before server configuration has been retrieved.
netinstrumentation to use shim api.
- Converted child_process instrumentation to newer shim style.
- Converted Timers instrumentation to newer shim style.
- Converted zlib instrumentation to use shim API.
codec.decode()callback bug that would re-call a callback with an error thrown within the callback.
Fixed bug requiring Cross Application Tracing (CAT) to be enabled for Distributed Tracing (DT)
acceptDistributedTracePayloadAPIs to function. DT configuration will no longer consider CAT configuration.
Fixed bug in wrap() that would fail to wrap callbacks if the callback index was 0.
The agent now respects
Previously, this environment variable was stored as a string. The environment variable is now stored as a float.
- Converted File System instrumentation to use newer shim style.
- Converted DNS instrumentation to newer shim style.
- Added tracking of callbacks to DNS instrumentation.
- Converted crypto instrumentation to newer shim style.
- Updated domains instrumentation to use an instrumentation shim.
- Refactored the global instrumentation to use the shim API.
- Ported inspector instrumentation to use an instrumentation shim.
- Ported async_hooks based promise instrumentation over to using shims.
- Added shim types for core instrumentations.
- Improved benchmark comparison output.
Agent instrumentation will no longer interfere with promisification of core methods.
Some core methods expose pre-promisified versions of the methods as a reference on the method itself. When instrumenting these methods, it neglected to forward these references onto the wrapper function. Now the instrumentation will properly forward property look ups to the original method.
Fixed outbound https call to use example.com to resolve integration test issue.
Fixed tests for ioredis 4.0.0 and above.
Implemented enforcement of
Any payload during the harvest sequence that exceeds the configured limit will be discarded.
Added JSON-formatted output to benchmarks to enable automated benchmark comparison.
Updated the benchmark runner to measure specifically userland CPU overhead.
Added DatastoreShim benchmarks.
Fixed MongoDB instrumentation for driver versions greater than 3.0.6.
Mongo 3.0.6 removed metadata the Agent relied upon to instrument the driver. This fixes that by going back to the old method of manually listing all objects and methods to instrument.
Updated MySQL versioned tests to run against the latest release.
Added support for distributed tracing.
Distributed tracing lets you see the path that a request takes as it travels through your distributed system. By showing the distributed activity through a unified view, you can troubleshoot and understand a complex system better than ever before.
Distributed tracing is available with an APM Pro or equivalent subscription. To see a complete distributed trace, you need to enable the feature on a set of neighboring services. Enabling distributed tracing changes the behavior of some New Relic features, so carefully consult the transition guide before you enable this feature.
To enable distributed tracing, set
newrelic.jsfile, or set
NEW_RELIC_DISTRIBUTED_TRACING_ENABLEDin your environment.
Added a warning for too-new versions of Node.js during agent startup.
Appropriately obfuscated SQL statements will now be included in all transaction traces.
Previously, the agent would only include the SQL statements if the corresponding query was sufficiently slow.
Added ability to execute instrumentation functions in the context of the segment the segment descriptor is describing.
record*methods supplied by all instrumentation shim classes now allow for a function to be executed under the context of the segment the record call will produce. This may be done by supplying a function in the
inContextkey for the segment descriptor passed to the record method.
Reservoirs will now respect setting their size to 0.