resolve resolveip command not found issue in integration test

[freyayunfu]

Fixes (partially) #7842 .

Description

We use resolveip to get the hose ip address when generating the keys and certs, which command not necessary already installed/worked well on that platform where user run the integration test. This would cause the integration test fail. Replacing resolveip with curl ifconfig.me to get the host ip address will be more robust and prevent the integration test fail.

However, I'm still getting below error after this issue is resolved: java.net.ConnectException: Connection refused: /192.168.99.100:8081 Coordinator does not start up in time for some reason, which requires more research.

[leventov]

Now all integration tests fail in CI, e. g. see https://travis-ci.org/apache/incubator-druid/jobs/597175561

In this PR: #6702 I've experimented with another approach to solving this, like

if [[ "$OSTYPE" == "darwin"* ]]; then
  # On Mac OS X resolveip may not be available
  export DOCKER_HOST_IP=$(dig +short $HOSTNAME | awk '{ print ; exit }')
else
  export DOCKER_HOST_IP=$(resolveip -s $HOSTNAME)
fi

See changes to Dockerfile and some other files but without much success, either

[freyayunfu]

Now all integration tests fail in CI, e. g. see https://travis-ci.org/apache/incubator-druid/jobs/597175561

In this PR: #6702 I've experimented with another approach to solving this, like

if [[ "$OSTYPE" == "darwin"* ]]; then
  # On Mac OS X resolveip may not be available
  export DOCKER_HOST_IP=$(dig +short $HOSTNAME | awk '{ print ; exit }')
else
  export DOCKER_HOST_IP=$(resolveip -s $HOSTNAME)
fi

See changes to Dockerfile and some other files but without much success, either

Thanks @leventov But before we do the PR for selfDiscovered endpoint, do we know what is the reason for the connection refused for coordinator in integration test? Is it because we didn't wait for enough time so that zookeeper hasn't added the node to cluster, or something else? I mean, I don't think PR #6702 will resolve the issue before we confirm the reason why the integration test failed.

[leventov]

@woshifudayun I don't know the reasons. Otherwise, that PR wouldn't be stuck.

[gianm]

This needs an internet connection, and relies on an external web service that might go away. I think we can do it without needing one. How about:

host `hostname` | perl -nle '/has address (.*)/ && print $1'

The two most common dns lookup commands AFAIK are host and dig, so probably either one would be fine.

[freyayunfu]

Agree. And i think @leventov already had a fix based on dig here: #6702 which also looks good to me.

[suneet-s]

@woshifudayun Thanks for your fix! I used this as inspiration to get myself unblocked and submitted a PR here - #9065

Would love if you could look over this change!

[stale]

This pull request has been marked as stale due to 60 days of inactivity. It will be closed in 4 weeks if no further activity occurs. If you think that's incorrect or this pull request should instead be reviewed, please simply write any comment. Even if closed, you can still revive the PR at any time or discuss it on the dev@druid.apache.org list. Thank you for your contributions.

[stale]

This pull request/issue has been closed due to lack of activity. If you think that is incorrect, or the pull request requires review, you can revive the PR at any time.