Investigate flaky parallel/test-net-better-error-messages-port-hostname

[Trott]

• Version: v8.0.0-pre
• Platform: centos5-32
• Subsystem: test, net

https://ci.nodejs.org/job/node-test-commit-linux/8716/nodes=centos5-32/console

not ok 740 parallel/test-net-better-error-messages-port-hostname
 ---
 duration_ms: 40.342
 severity: fail
 stack: |-
 
 assert.js:81
 throw new assert.AssertionError({
 ^
 AssertionError: 'EAI_AGAIN' === 'ENOTFOUND'
 at Socket.<anonymous> (/home/iojs/build/workspace/node-test-commit-linux/nodes/centos5-32/test/parallel/test-net-better-error-messages-port-hostname.js:11:10)
 at Socket.<anonymous> (/home/iojs/build/workspace/node-test-commit-linux/nodes/centos5-32/test/common.js:461:15)
 at emitOne (events.js:115:13)
 at Socket.emit (events.js:210:7)
 at connectErrorNT (net.js:1052:8)
 at _combinedTickCallback (internal/process/next_tick.js:80:11)
 at process._tickCallback (internal/process/next_tick.js:104:9)
 ...

[Trott]

@nodejs/testing

[Trott]

EAI_AGAIN is basically "DNS timed out" which seems like it could be legitimate. I'm not convinced thiat we shouldn't allow for that. Thoughts?

[Trott]

I could be mistaken, but it looks to me that test-digitalocean-centos5-x86-1 is configured to use Google DNS. So this test reaches out to the Internet. I wonder if we should move it to test/internet (although I suspect an awful lot of our tests do DNS lookups like this so maybe just live with it and allow for the occasional EAI_AGAIN). @nodejs/build

[cjihrig]

The test could be updated to allow for EAI_AGAIN. I wonder why EAI_AGAIN wasn't clumped into the group of errors comprising the made up ENOTFOUND (not saying we should do that now).

Another alternative would be to monkey patch the DNS to always return the correct error. I don't think that's a great idea, but I don't like moving things to test/internet either.

[Trott]

@addaleax Any chance this is a good candidate for the test-double DNS server that always returns an error that you created in common?

[addaleax]

@Trott I think that’s doable and makes sense, yes. :) I can look into it, but if you want this could actually be a nice second or third contribution or so.

Edit: That might also actually allow us to test a real domain + port combination instead of the fake *** one here

[addaleax]

Sorry, I misread that, net uses dns.lookup, not dns.resolve, so using dns.setServers() won’t have any effect here.

[BridgeAR]

This got fixed by 0309619