[
https://issues.apache.org/jira/browse/SOLR-10898?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Hoss Man updated SOLR-10898:
----------------------------
Attachment: SOLR-10898.patch
Here's a patch that keeps the test around, but reworks it so it instead of a
fixed number of requests, it executes them in a loop until each of the (2)
replicas has received at least one request -- or until a total of 1000 requests
have been made, at which point the assertion failure message is clear that the
problem could just be a *very* unlucky seed.
> TestRandomRequestDistribution.testRequestTracking is terrible and should be
> removed
> -----------------------------------------------------------------------------------
>
> Key: SOLR-10898
> URL: https://issues.apache.org/jira/browse/SOLR-10898
> Project: Solr
> Issue Type: Bug
> Security Level: Public(Default Security Level. Issues are Public)
> Reporter: Hoss Man
> Attachments: SOLR-10898.patch
>
>
> TestRandomRequestDistribution.testRequestTracking() is a badly written test
> that tries to assert that if 10 requests are sent to a shard has 2 replicas,
> at least 1 of those requests must be routed to each replica.
> This is essentially a test that says "flip a coin 10 times, fail if all 10
> are heads *OR* if all 10 are tails"
> Statistically speaking 2 out of every 1024 runs of this test (1/512) should
> fail.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
