[
https://issues.apache.org/jira/browse/HBASE-13078?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14492812#comment-14492812
]
Josh Elser commented on HBASE-13078:
------------------------------------
Thanks for taking a look, Nick.
bq. a smoke test like this one that can be run to validate tracing is
configured correctly cluster-wide (ie, and spans are coming out of both hbase
and hdfs). Maybe the latter is unreasonable?
Difficult, perhaps. The instrumentation in HDFS will (continue to) be a moving
target, I imagine. It might be better to get good tests in HDFS to ensure that
tracing occurs where desired. If we can guarantee that spans in HDFS make it to
the SpanReceiver, we shouldn't need to retest here in HBase.
bq. a medium or large test that verifies a single traced invocation with
everything enabled results in spans logged (and perhaps some validation thereof)
Cool, I'll open up another issue for this. Opinions/thoughts on what this would
look like are suggested/welcome. At a minimum, I can peruse the usages and try
to tailor a test around that.
> IntegrationTestSendTraceRequests is a noop
> ------------------------------------------
>
> Key: HBASE-13078
> URL: https://issues.apache.org/jira/browse/HBASE-13078
> Project: HBase
> Issue Type: Test
> Components: integration tests
> Reporter: Nick Dimiduk
> Assignee: Josh Elser
> Priority: Critical
> Fix For: 2.0.0, 1.1.0, 0.98.13, 1.0.2
>
> Attachments: HBASE-13078-0.98-removal.patch,
> HBASE-13078-0.98-v1.patch, HBASE-13078-v1.patch, HBASE-13078.patch
>
>
> While pair-debugging with [~jeffreyz] on HBASE-13077, we noticed that
> IntegrationTestSendTraceRequests doesn't actually assert anything. This test
> should be converted to use a mini cluster, setup a POJOSpanReceiver, and then
> verify the spans collected.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
