[Hadoop-common-dev] Review Request: HADOOP-7104 Remove unnecessary DNS reverse lookups from RPC layer

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/316/#review106
-----------------------------------------------------------



trunk/src/test/core/org/apache/hadoop/security/TestSecurityUtil.java
<https://reviews.apache.org/r/316/#comment217>

I think this test case could be improved a few ways:

- Rather than actually rely on forward/reverse DNS for the localhost matching up in the test case, you can do something like:

private InetAddress mockAddr(String reverseTo) {
InetAddress mock = Mockito.mock(InetAddress.class);
doReturn(reverseTo).when(mock).getCanonicalHostName();
return mock;
}

then for the test:

verify("hdfs/_HOST@REALM", mockAddr("foohost"), "hdfs/foohost@REALM");

- another useful test to make sure reverse DNS doesn't happen:

InetAddress mock = Mockito.mock(InetAddress.class);
verify("hdfs/specifiedhost@REALM", mock, "hdfs/specifiedhost@REALM");
Mockito.verify(mock, Mockito.never()).getCanonicalHostName();

This should hopefully make the tests less fussy on hudson boxes where experience shows DNS is often screwy.

- Need to test the newly allowed case for hdfs@REALM instead of hdfs/...@REALM, right?


- Todd

Hey Todd,

We were discussing redirecting these to the jira itself. Do you know
what happened?

I'm finding it really hard to follow both jira and these. Is there
an INFRA ticket you know about?

thanks,
Arun

Hey Arun,

Sorry, this one snuck by my filters and ended up in a weird place in gmail
:)

The relevant JIRA is INFRA-3153. I've been remiss in working with the infra
guys on this one over the holiday season. I just pinged Gavin and hopefully
we'll get to this soon together.

-Todd