| Age | Commit message (Collapse) | Author |
|---|
|
* Exporter/OcAgent: Add a fake impl of TraceService.
* Move the fake server impl to test.
|
|
* Add dropwizard metrics producer
* Use Summary Type
* Fix reviews
* Fix reviews
|
|
* Exporter/OCAgent: Add OcAgentNodeUtils.
Add utilities for detecting and creating Node.
Equivalent to
https://github.com/census-ecosystem/opencensus-go-exporter-ocagent/blob/master/nodeinfo.go.
* Fix nullness checker.
* Use a local VERSION string in OC-Agent Exporter.
|
|
|
|
|
|
* Contrib: Add opencensus-contrib-opencensus-proto-util.
* Update CHANGELOG.
|
|
(#1426)" (#1432)
This reverts commit d3fa0e5c60903f95c184d7c7bafae8fdd12156ad.
|
|
This commit improves the thread safety of OpenCensusTraceContextDataInjector by
following the thread safety requirements in the Javadocs of the overridden
methods from ContextDataInjector. It also handles the possibility of
ThreadContext.getThreadContextMap() returning null by adding a Nullable
annotation in a Checker Framework stub file and adding null checks to the code.
|
|
The "metrics" package isn't ready to be released yet, so this commit moves it
out of the API artifact. The package can still be accessed by the stats
implementation in impl_core/. This commit can be reverted once the package is
ready to be exposed. The moved package names also contain "temporary" so that
there is no possibility of class name conflicts between different versions of
opencensus-api and opencensus-impl-core. For example,
io.opencensus.metrics.export is renamed to
io.opencensus.implcore.temporary.metrics.export.
|
|
This commit adds the internal package to import-control.xml and removes the use
of org.apache.logging.log4j.core.impl.ThreadContextDataInjector.
|
|
log4j2. (#1411)
This commit renames the artifact to opencensus-contrib-log-correlation-log4j2
and renames the package to io.opencensus.contrib.logcorrelation.log4j2. The
name change emphasizes that the artifact only works with Log4j 2 and allows us
to add support for other versions of Log4j in the future.
|
|
* Initial import of OpenCensus/Sleuth integration.
* Minor fixes for integration with master.
* Adds a README.
* Removes hardcoded dependencies in build file.
* Adds contrib to spring_sleuth file path.
* Updates package names appropriately based on file renames.
* Fixes build paths.
* Fixes for build checks.
* Mark as experimental.
* Minor fixes for build files (space vs tab).
* Minor fixes for build files (space vs tab).
* Update README to absolve Spring Sleuth team of any responsibility for this spring_sleuth plugin.
* Minor fixes for import control (space vs tab).
* Tag public classes @since 0.16
* Adds javadoc for public methods not already documented.
* Updates OpenCensusSleuthTracer to pass nullness checks.
* Fixes for formatting issues.
* Fixes for typos.
* Suppresses the check return value warning added by gRPC.
* Suppresses the check return value warning added by gRPC.
|
|
* Remove guava dependency from the API artifact.
* Make LowerCaseBase16Encoding static and remove import control.
|
|
This commit takes the approach of implementing Log4j's ContextDataInjector. It
inserts three keys into each log event: "`openCensusTraceId`", "`openCensusSpanId`",
and "`openCensusTraceSampled`". The tracing data can be accessed with
LogEvent.getContextData() or included in a layout. See
https://logging.apache.org/log4j/2.x/log4j-core/apidocs/org/apache/logging/log4j/core/ContextDataInjector.html.
Here is an example of a pattern layout that includes all three tracing fields
and a log entry that could be produced by the layout:
Pattern layout:
"`%d{HH:mm:ss.SSS} [%t] traceId=%X{openCensusTraceId} spanId=%X{openCensusSpanId} sampled=%X{openCensusTraceSampled} %-5level %logger{36} - %msg%n`"
Log entry:
"`20:45:52.929 [main] traceId=0af7a7bef890695f1c5e85a8e7290164 spanId=d3f07c467ec2fbb2 sampled=true WARN io.opencensus.contrib.logcorrelation.log4j.demo.OpenCensusLog4jLogCorrelationDemo - my log message`"
The ContextDataInjector implementation must be specified with a system property,
i.e.,
"`-Dlog4j2.contextDataInjector=io.opencensus.contrib.logcorrelation.log4j.OpenCensusTraceContextDataInjector`".
The system property
"`io.opencensus.contrib.logcorrelation.log4j.OpenCensusTraceContextDataInjector.spanSelection`"
controls whether tracing data is added to all spans, no spans, or only sampled spans.
|
|
This commit also merges new changes to checkstyle.xml and fixes new checkstyle
warnings related to Javadocs.
|
|
* Stats: Implement MetricProducer in Stats.
* Add unit tests.
* Remove MetricProduce from Stats API. Register it statically in the impl.
|
|
|
|
* Metrics: Implement ExportComponent, make ProducerManager abstract.
* Metrics: Add impl and impllite.
* Resolve two TODOs.
* Fix import path, leave a TODO for getMetricRegistry.
* Make NoopMetricProducerManager private. Add missing since tag.
|
|
Support recording `Metric`s in stats impl. This works as the following:
- When a `View` is registered, convert that `View` to a `MetricDescriptor` and register it;
- When a `Measurement` along with a `TagContext` are recorded, convert them into a data row consisting of `LabelValue`s and `Point`s;
- If `getMetrics()` is called, convert the `MetricDescriptor`s and data rows into `Metric`s. Then flush all data rows.
This PR only contains internal support in the impl. None of these changes are user-visible yet.
|
|
709d97aa321d5729988fd63b960bbece04cfba10 modified the -AskipDefs argument to the
Checker Framework (a regular expression) in a way that caused it to skip
checking all files. This commit fixes the regular expression and the new
Checker Framework warnings.
|
|
|
|
|
|
* Add metric components for export and main package and metric producer.
* Solve first round of comments.
* Minor comments fixes.
|
|
|
|
* Stats: Add attachment key for TraceId and SpanId.
* Move the keys to an independent artifact and add methods for encoding.
* Add this change to CHANGELOG.
* Update description and tests.
* Have one method that takes SpanContext instead of two.
* Rename method to putSpanContextAttachments.
|
|
Start adding log correlation for Stackdriver Logging.
|
|
The demo project will be moved to
https://github.com/census-ecosystem/opencensus-experiments.
|
|
|
|
* Stackdriver Trace Exporter: add resource labels to Spans.
* Fix review comments.
* Make resource labels static and only initialize once.
* Remove SuppressWarnings("unchecked")
* Passing resource_label map to generateSpan().
* Fix a checkstyle error.
* Add more unit tests on resource labels.
|
|
|
|
* Move Metrics to a separate artifact.
* Fix review comments.
* Update one TODO about the naming of the artifact.
|
|
|
|
|
|
|
|
- Each task is treated as separate job to run in parallel.
- single script linux_presumbit.sh performs all task based TASK env.
|
|
Add kokoro build scripts for osx platform.
|
|
* compiled on osx.
|
|
* compiled buildscripts/kokoro/windows.bat on Windows instance.
|
|
|
|
Move monitored resource detection utils to a separate artifact, so that it can be reused by both Stackdriver Monitoring and Stackdriver Trace.
|
|
|
|
This commit uses the Checkstyle ImportControl feature
(http://checkstyle.sourceforge.net/config_imports.html#ImportControl) to specify
the allowed imports for each package. Specifying imports has several benefits:
- It makes some parts of the design more concrete. For example, it shows that
'stats' can depend on 'tags', but 'tags' shouldn't depend on 'stats' or
'trace'.
- We can use it to restrict access to 'internal' packages.
- It can help us remove the dependency on Guava incrementally. For example,
this commit specifies that the API can only depend on
com.google.common.io.BaseEncoding, which should help us avoid accidentally
adding more uses of Guava to the API.
One downside is that the check seems to only apply to import statements, not
qualified class names.
This commit specifies imports positively, using "allow" XML elements. If the
lists of imports are too difficult to maintain, especially for packages such as
exporters that have many imports, we could switch to using "disallow" elements
in some places.
|
|
|
|
|
