New Features
Added new methods to expose
Transaction
details:Transaction.GetTraceMetadata()
returns a TraceMetadata which contains distributed tracing identifiers.Transaction.GetLinkingMetadata()
returns a LinkingMetadata which contains the fields needed to link data to a trace or entity.
Added a new plugin for the Logrus logging framework with the new _integrations/logcontext/nrlogrusplugin package. This plugin leverages the new
GetTraceMetadata
andGetLinkingMetadata
above to decorate logs.To enable, set your log's formatter to the
nrlogrusplugin.ContextFormatter{}
logger := logrus.New()logger.SetFormatter(nrlogrusplugin.ContextFormatter{})The logger will now look for a
newrelic.Transaction
inside its context and decorate logs accordingly. Therefore, the Transaction must be added to the context and passed to the logger. For example, this logging calllogger.Info("Hello New Relic!")must be transformed to include the context, such as:
ctx := newrelic.NewContext(context.Background(), txn)logger.WithContext(ctx).Info("Hello New Relic!")Added support for NATS and NATS Streaming monitoring with the new _integrations/nrnats and _integrations/nrstan packages. These packages support instrumentation of publishers and subscribers.
Enables ability to migrate to Configurable Security Policies (CSP) on a per agent basis for accounts already using High Security Mode (HSM).
- Previously, if CSP was configured for an account, New Relic would not allow an agent to connect without the
security_policies_token
. This led to agents not being able to connect during the period between when CSP was enabled for an account and when each agent is configured with the correct token. - With this change, when both HSM and CSP are enabled for an account, an agent (this version or later) can successfully connect with either
high_security: true
or the appropriatesecurity_policies_token
configured - allowing the agent to continue to connect after CSP is configured on the account but before the appropriatesecurity_policies_token
is configured for each agent.
- Previously, if CSP was configured for an account, New Relic would not allow an agent to connect without the
New Features
Added support for Micro monitoring with the new _integrations/nrmicro package. This package supports instrumentation for servers, clients, publishers, and subscribers.
Added support for creating static
WebRequest
instances manually via theNewStaticWebRequest
function. This can be useful when you want to create a web transaction but don't have anhttp.Request
object. Here's an example of creating a staticWebRequest
and using it to mark a transaction as a web transaction:hdrs := http.Headers{}u, _ := url.Parse("http://example.com")webReq := newrelic.NewStaticWebRequest(hdrs, u, "GET", newrelic.TransportHTTP)txn := app.StartTransaction("My-Transaction", nil, nil)txn.SetWebRequest(webReq)
2.10.0
New Features
Added support for custom events when using nrlambda. Example Lambda handler which creates custom event:
func handler(ctx context.Context) {if txn := newrelic.FromContext(ctx); nil != txn {txn.Application().RecordCustomEvent("myEvent", map[string]interface{}{"zip": "zap",})}fmt.Println("hello world!")}
2.9.0
New Features
Added support for gRPC monitoring with the new _integrations/nrgrpc package. This package supports instrumentation for servers and clients.
Added new ExternalSegment fields
Host
,Procedure
, andLibrary
. These optional fields are automatically populated from the segment'sURL
orRequest
if unset. Use them if you don't have access to a request or URL but still want useful external metrics, transaction segment attributes, and span attributes.Host
is used for external metrics, transaction trace segment names, and span event names. The host of segment'sRequest
orURL
is the default.Procedure
is used for transaction breakdown metrics. If set, it should be set to the remote procedure being called. The HTTP method of the segment'sRequest
is the default.Library
is used for external metrics and the"component"
span attribute. If set, it should be set to the framework making the call."http"
is the default.
With the addition of these new fields, external transaction breakdown metrics are changed:
External/myhost.com/all
will now report asExternal/myhost.com/http/GET
(provided the HTTP method isGET
).HTTP Response codes below
100
, except0
and5
, are now recorded as errors. This is to supportgRPC
status codes. If you start seeing new status code errors that you would like to ignore, add them toConfig.ErrorCollector.IgnoreStatusCodes
or your server side configuration settings.Improve logrus support by introducing nrlogrus.Transform, a function which allows you to turn a logrus.Logger instance into a newrelic.Logger. Example use:
l := logrus.New()l.SetLevel(logrus.DebugLevel)cfg := newrelic.NewConfig("Your Application Name", "__YOUR_NEW_RELIC_LICENSE_KEY__")cfg.Logger = nrlogrus.Transform(l)As a result of this change, the nrlogrus package requires logrus version
v1.1.0
and above.
2.8.1
Bug Fixes
- Removed
nrmysql.NewConnector
since go-sql-driver/mysql has not yet releasedmysql.NewConnector
.
2.8.0
New Features
Support for Real Time Streaming
- Event data is now sent to New Relic every five seconds, instead of every minute. As a result, transaction, error, and custom events will now be available in New Relic dashboards in near real time. For more information on how to view your events with a five-second refresh, see the real time streaming documentation.
- Note that the overall limits on how many events can be sent per minute have not changed. Also, span events, metrics, and trace data is unaffected, and will still be sent every minute.
Introduce support for databases using database/sql. This new functionality allows you to instrument MySQL, PostgreSQL, and SQLite calls without manually creating DatastoreSegments.
Database Library Supported | Integration Package |
---|---|
Using these database integration packages is easy! First replace the driver with our integration version:
import ( // import our integration package _ "github.com/newrelic/go-agent/_integrations/nrmysql" )
func main() { // open "nrmysql" in place of "mysql" db, err := sql.Open("nrmysql", "user@unix(/path/to/socket)/dbname") }
Second, use the ExecContext
, QueryContext
, and QueryRowContext
methods of sql.DB, sql.Conn, sql.Tx, and sql.Stmt and provide a transaction-containing context. Calls to Exec
, Query
, and QueryRow
do not get instrumented.
ctx := newrelic.NewContext(context.Background(), txn) row := db.QueryRowContext(ctx, "SELECT count(*) from tables")
If you are using a database/sql database not listed above, you can write your own instrumentation for it using InstrumentSQLConnector, InstrumentSQLDriver, and SQLDriverSegmentBuilder. The integration packages act as examples of how to do this.
For more information, see the Go agent documentation on instrumenting datastore segments.
Bug Fixes
- The http.RoundTripper returned by NewRoundTripper no longer modifies the request. Our thanks to @jlordiales for the contribution.
2.7.0
New Features
Added support for server side configuration. Server side configuration allows you to set the following configuration settings in the New Relic APM UI:
Config.TransactionTracer.Enabled
Config.ErrorCollector.Enabled
Config.CrossApplicationTracer.Enabled
Config.TransactionTracer.Threshold
Config.TransactionTracer.StackTraceThreshold
Config.ErrorCollector.IgnoreStatusCodes
For more information see the server side configuration documentation.
Added support for AWS Lambda functions in the new nrlambda package. Please email lambda_preview@newrelic.com if you are interested in learning more or previewing New Relic Lambda monitoring. This instrumentation package requires
aws-lambda-go
version v1.9.0 or higher.
2.6.0
New Features
Added support for async: the ability to instrument multiple concurrent goroutines, or goroutines that access or manipulate the same Transaction.
The new
Transaction.NewGoroutine() Transaction
method allows transactions to create segments in multiple goroutines!NewGoroutine
returns a new reference to theTransaction
. This must be called any time you are passing theTransaction
to another goroutine which makes segments. Each segment-creating goroutine must have its ownTransaction
reference. It does not matter if you call this before or after the other goroutine has started.All
Transaction
methods can be used in anyTransaction
reference. TheTransaction
will end whenEnd()
is called in any goroutine.Example passing a new
Transaction
reference directly to another goroutine:go func(txn newrelic.Transaction) {defer newrelic.StartSegment(txn, "async").End()time.Sleep(100 * time.Millisecond)}(txn.NewGoroutine())Example passing a new
Transaction
reference on a channel to another goroutine:ch := make(chan newrelic.Transaction)go func() {txn := <-chdefer newrelic.StartSegment(txn, "async").End()time.Sleep(100 * time.Millisecond)}()ch <- txn.NewGoroutine()Added integration support for
aws-sdk-go
andaws-sdk-go-v2
.When using these SDKs, a segment will be created for each out going request. For DynamoDB calls, these will be Datastore segments and for all others they will be External segments.
- v1 Documentation
- v2 Documentation
- Added span event and transaction trace segment attribute configuration. You may control which attributes are captured in span events and transaction trace segments using the
Config.SpanEvents.Attributes
andConfig.TransactionTracer.Segments.Attributes
settings. For example, if you want to disable the collection of"db.statement"
in your span events, modify your config like this:
cfg.SpanEvents.Attributes.Exclude = append(cfg.SpanEvents.Attributes.Exclude,newrelic.SpanAttributeDBStatement)To disable the collection of all attributes from your transaction trace segments, modify your config like this:
cfg.TransactionTracer.Segments.Attributes.Enabled = false
Bug Fixes
- Fixed a bug that would prevent External Segments from being created under certain error conditions related to Cross Application Tracing.
Miscellaneous
Improved linking between Cross Application Transaction Traces in the APM UI. When
Config.CrossApplicationTracer.Enabled = true
, External segments in the Transaction Traces details will now link to the downstream Transaction Trace if there is one. Additionally, the segment name will now include the name of the downstream application and the name of the downstream transaction.Update attribute names of Datastore and External segments on Transaction Traces to be in line with attribute names on Spans. Specifically:
"uri"
=>"http.url"
"query"
=>"db.statement"
"database_name"
=>"db.instance"
"host"
=>"peer.hostname"
"port_path_or_id"
+"host"
=>"peer.address"
2.5.0
- Added support for New Relic Browser using the new
BrowserTimingHeader
method on theTransaction
which returns a BrowserTimingHeader. The New Relic Browser JavaScript code measures page load timing, also known as real user monitoring. The Pro version of this feature measures AJAX requests, single-page applications, JavaScript errors, and much more! Example use:
func browser(w http.ResponseWriter, r *http.Request) { hdr, err := w.(newrelic.Transaction).BrowserTimingHeader() if nil != err { log.Printf("unable to create browser timing header: %v", err) } // BrowserTimingHeader() will always return a header whose methods can // be safely called. if js := hdr.WithTags(); js != nil { w.Write(js) } io.WriteString(w, "browser header page")}
- The Go agent now collects an attribute named
request.uri
on Transaction Traces, Transaction Events, Error Traces, and Error Events.request.uri
will never contain user, password, query parameters, or fragment. To prevent the request's URL from being collected in any data, modify yourConfig
like this:
cfg.Attributes.Exclude = append(cfg.Attributes.Exclude, newrelic.AttributeRequestURI)
2.4.0 Notes
- Introduced
Transaction.Application
method which returns theApplication
that started theTransaction
. This method is useful since it may prevent having to pass theApplication
to code that already has access to theTransaction
. Example use:
txn.Application().RecordCustomEvent("customerOrder", map[string]interface{}{ "numItems": 2, "totalPrice": 13.75,})
- The
Transaction.AddAttribute
method no longer acceptsnil
values since our backend ignores them.