This has been consistently happening for the oojs-core-npm gate job for several days now:
https://integration.wikimedia.org/ci/job/oojs-core-npm/449/console
00:01:06.099 Running "karma:ci1" (karma) task 00:01:06.108 INFO [karma]: Karma v0.12.31 server started at http://localhost:9876/ 00:01:06.113 INFO [launcher]: Starting browser chrome on SauceLabs 00:01:06.115 INFO [launcher]: Starting browser firefox on SauceLabs 00:01:06.117 INFO [launcher]: Starting browser internet explorer 11 (Windows 7) on SauceLabs 00:01:08.166 ERROR [launcher.sauce]: Can not start chrome 00:01:08.589 Failed to start Sauce Connect: 00:01:08.589 Error: GET https://saucelabs.com/rest/v1/oojs/tunnels?full=1: Couldn't resolve host name. 00:01:08.589 ERROR [launcher.sauce]: Can not start firefox 00:01:08.590 Failed to start Sauce Connect: 00:01:08.590 Error: GET https://saucelabs.com/rest/v1/oojs/tunnels?full=1: Couldn't resolve host name. 00:01:08.590 ERROR [launcher.sauce]: Can not start internet explorer 11 (Windows 7) 00:01:08.591 Failed to start Sauce Connect: 00:01:08.591 Error: GET https://saucelabs.com/rest/v1/oojs/tunnels?full=1: Couldn't resolve host name. 00:01:08.591 Warning: Task "karma:ci1" failed. Use --force to continue. 00:01:08.591 00:01:08.591 Aborted due to warnings.
I was able to consistently reproduce this when running grunt ci on latest master of oojs-core via jenkins-deploy@integration-slave1401:/mnt/jenkins-workspace/workspace/oojs-core-npm.
However it's working fine from localhost for @Jdforrester-WMF and myself.
Did something change recently in the firewall or DNS configuration of WMFLabs or Eqiad that might cause this?
It last worked on Feb 24. This first failed on March 6.