v4.6.0: test failure, servers not starting?

[CodeMan99]

• Version: v4.6.0
• Platform: Ubuntu Linux

Build steps:

1. git checkout v4.6.0
2. ./configure --prefix="$HOME/.local" --dest-cpu=x64 --dest-os=linux
3. make
4. make test

/usr/bin/python tools/test.py --mode=release doctool message pseudo-tty parallel sequential -J
=== release test-https-connect-address-family ===
Path: parallel/test-https-connect-address-family
events.js:141
      throw er; // Unhandled 'error' event
      ^

Error: getaddrinfo ENOTFOUND localhost localhost:33215
    at errnoException (dns.js:27:10)
    at GetAddrInfoReqWrap.onlookup [as oncomplete] (dns.js:78:26)
Command: out/Release/node /home/code_m/dev/node/test/parallel/test-https-connect-address-family.js
=== release test-tls-connect-address-family ===
Path: parallel/test-tls-connect-address-family
events.js:141
      throw er; // Unhandled 'error' event
      ^

Error: getaddrinfo ENOTFOUND localhost localhost:35246
    at errnoException (dns.js:27:10)
    at GetAddrInfoReqWrap.onlookup [as oncomplete] (dns.js:78:26)
Command: out/Release/node /home/code_m/dev/node/test/parallel/test-tls-connect-address-family.js
[03:12|% 100|+ 1047|-   2]: Done
Makefile:114: recipe for target 'test' failed
make: *** [test] Error 1

Would not be surprised if this is a network failure on my end. I will dig in more after work today.

[MylesBorins]

@CodeMan99 are you running any firewalls? are you seeing the same failures if you manually compile and test v4.5.0?

[CodeMan99]

No firewalls that I set up.

Interestingly, v4.5.0 does fail in the same way. That was not the case when I built v4.5.0 when the release occurred.

Really unsure how my environment changed. I do not recall doing any system updates. I am not running any other software, and definitely not anything that should be using the extremely high port range.

[MylesBorins]

That is indeed quite strange. I'm going to close this issue for now as we know at the very least that this is not specific to v4.6.0. Please continue to update us on what is going on here, I'm interested in finding out what the cause is.

Also please feel free to ask any questions that may help you figure it out. Have you introduced anything like fail2ban recently?

[bnoordhuis]

Those tests expect that 'localhost' resolves to '::1', i.e., that the local DNS resolver has at least one AAAA record for localhost.

Some fixes have been made to make the tests more resilient (#7605, #7766) but they haven't been back-ported to a v4.x release yet.

[MylesBorins]

@bnoordhuis thanks for the insight! Both of those PR's have been backported to staging

[CodeMan99]

@thealphanerd Definitely just flakes. The tests passed today just fine. Only difference from yesterday is my physical location/network.

[MylesBorins]

a++ thanks for checking in