Fix flaky client metrics test #2170
Conversation
Signed-off-by: Yuri Shkuro <ys@uber.com>
Signed-off-by: Yuri Shkuro <ys@uber.com>
Signed-off-by: Yuri Shkuro <ys@uber.com>
Codecov Report
@@ Coverage Diff @@
## master #2170 +/- ##
==========================================
+ Coverage 96.14% 96.16% +0.01%
==========================================
Files 219 219
Lines 10585 10585
==========================================
+ Hits 10177 10179 +2
+ Misses 353 351 -2
Partials 55 55
Continue to review full report at Codecov.
|
| time.Sleep(500 * time.Microsecond) | ||
|
|
||
| t.Run("detect new client", func(t *testing.T) { | ||
| assert.EqualValues(t, 0, getGauge(), "start with gauge=0") |
There was a problem hiding this comment.
Why is this needed? What else could set the gauge to a non zero value earlier?
| params := ClientMetricsReporterParams{ | ||
| ExpireFrequency: 100 * time.Microsecond, | ||
| ExpireTTL: 5 * time.Millisecond, | ||
| ExpireFrequency: 1 * time.Millisecond, |
There was a problem hiding this comment.
can simply be time.Millisecond
| require.EqualValues(t, 1, gauge) | ||
| }) | ||
|
|
||
| t.Run("detect stale client", func(t *testing.T) { |
There was a problem hiding this comment.
Is it difficult to start with gauges at a specific value instead of relying on EmitBatch?
There was a problem hiding this comment.
I am not sure, it would probably require some backdoor. This way we're testing cleanly via the official APIs.
Resolves #2108
Supersedes #2168