The test assumes the threads would schedule in particular way, but they don't.
But what we really care to check is that we thread in the proper RL and it works.
We don't need to check that underlying queue impl works, that's done in its own tests.
So just verify these two things.
* Remove key in tags to reduce metrics count
Issue: https://github.com/knative/serving/issues/8609
Signed-off-by: Lance Liu <xuliuxl@cn.ibm.com>
* remove tag key for OpenCensus
Signed-off-by: Lance Liu <xuliuxl@cn.ibm.com>
* do not record for empty metric config
* Revert "do not record for empty metric config"
This reverts commit 539a5e4dbb.
* add a comment
* fix typo
* fix tests
* revert
* revert tests
* revert
* fix conflicts
* one more test file
* do not record for empty metric config
* Revert "do not record for empty metric config"
This reverts commit 539a5e4dbb.
* use metricstest package for test
* Remove unused code.
* Use raw strings to avoid escaping.
* Remove unneeded type conversions.
* Preallocate slices where possible.
* Use semantic equality in psbinding reconciler.
* split and update
* bug fixed
* change the copyright year
* use fake metadata in tests
* exclude test coverage for gcp_metadata
* add record
* add tests
* revert 2019 and add a new var
* change documents
* address comment
* change line
* Use different reporting periods based on the metrics backend. Original implementation was setting this value to 1 min to accommodate the minimum allowed value for stackdriver, but that is a sub optimal behavior for Prometheus. The change also allows operator to override the default values.
* Fix the units of reconciler latency from nano seconds to milliseconds.
* Address PR comments.
* Add better logging for reconciliation.
* Remove unnecessary lock in the UT.
* Address PR comments.
rusde
Victor Agababov
Lance Liu
Yanwei Guo
Markus Thömmes
Matt Moore
Mustafa Demirhan