[SPARK-26554][BUILD][FOLLOWUP] Use GitHub instead of GitBox to check HEADER

[dongjoon-hyun]

What changes were proposed in this pull request?

This PR uses GitHub repository instead of GitBox because GitHub repo returns HTTP header status correctly.

How was this patch tested?

Manual.

$ ./do-release-docker.sh -d /tmp/test -n
Branch [branch-2.4]:
Current branch version is 2.4.1-SNAPSHOT.
Release [2.4.1]:
RC # [1]:
This is a dry run. Please confirm the ref that will be built for testing.
Ref [v2.4.1-rc1]:

[HyukjinKwon]

Hm, I wonder if I completely agree with it. Mixing gitbox and github could look a bit confusing. The previous ! curl -s --fail "$ASF_REPO_WEBUI;a=commit;h=$1" | grep '404 Not Found' > /dev/null is good enough I guess. I'm okay but let's don't do this if it's not only me feeling in this way.

[cloud-fan]

It's good enough but it's a hack to me. Using Github repo at this single place is also a hack for now, but according to #23476 (comment) , seems we can switch to Github repo completely in the future.

[HyukjinKwon]

If we're going to switch to github, I'm okie. Don't block by me on this. Either way is fine to me.

[srowen]

Either should be fine. It has just occurred to me though, what if some of us pushed to gitbox and some to github? how would sync resolve it? what rebases on what? I guess it is better to push to one consistently, and I presume most of us (all?) are pushing to github?

[dongjoon-hyun]

For now, I don't see a clear LGTM for this PR.
And, at least, I'm currently pushing GitBox only.
Please let me know if the decision is made.

[srowen]

Hm, I suppose I think we should consistently push to github. That was the advice I think we sent (?) to committers, and I can update the committer docs on this if any exist. At least we'd have a standard.

[dongjoon-hyun]

@srowen . To give you my background, I followed your email instruction exactly and hit the followings on the same day after I pushed to GitBox first time.

• https://lists.apache.org/thread.html/9c09b5b3678ccda5d2ad2180d7c6463ba0e84aeb7b218e6c03002035@%3Cdev.spark.apache.org%3E
• https://issues.apache.org/jira/browse/INFRA-17401

The above situation was the one I mentioned here: #23476 (comment)

For me, since last bad experience, I'm pushing to only GitBox. So, I'm not sure whether GitHub to GitBox sync is safe or not in these days.

That's the reason I've been careful and tried not to cause any sync issue. If everything has no problem, I have no problem to switch back to the standard day~.

[dongjoon-hyun]

BTW, until now, no one gave approval to this. If we prefer to GitBox, could you guys give +1 for this PR, too?

[srowen]

That sync has been resolved, right? or at least, we do need that to work going forward, rather than work around it. Is it a blocker right now?

Overall I think we should standardize on github.com remotes over gitbox, so I think this could change again to use all the github.com remotes. Is that OK with you? that would be OK with me.

[SparkQA]

Test build #100871 has finished for PR 23482 at commit d7a048b.

• This patch passes all tests.
• This patch merges cleanly.
• This patch adds no public classes.

[cloud-fan]

LGTM for this change. In the followup we can completely switch to github, e.g. update the committer doc, update other release scripts, etc.

[HyukjinKwon]

Okay, then it's good. LGTM too

[dongjoon-hyun]

Thank you, @cloud-fan and all.
Merged to master/branch-2.4.