Retry all HttpException & SocketException in to methods (workaround http -1)

[borgstrom]

This PR is the result of needing to work around https://issues.jenkins-ci.org/browse/JENKINS-45142

We were seeing the following error quite frequently when scanning all repos in our github org:

ERROR: Server returned HTTP response code: -1, message: 'null' for URL: https://api.github.com
java.net.SocketException: Socket closed

The retry code in parse was not working as expected, so this moves the retry code directly into the to methods.

[jglick]

Approach sounds fine to me, especially if it has been demonstrated to work better than the original fix. Awaiting tuning of status code handling.

[jglick]

Can this be limited to SocketTimeoutException, as in the original?

[borgstrom]

No, because HttpException extends from IOException.

I will limit the retry mechanism to only HttpExceptions with status codes >= 400 && SocketException-- I feel that there's merit to retrying other socket exceptions, such as connection reset errors.

[jglick]

This amends #359 FTR.

[borgstrom]

I have just uploaded a snapshot of github-api-plugin with this branch to my jenkins instance and kicked off a full org scan.

[borgstrom]

This implementation is not working as expected. We're still seeing -1 status HTTP responses in our logs periodically:

[Tue Aug 29 10:44:46 PDT 2017] Push event to branch github-reading in repository NerdWallet/museum UPDATED event from 192.30.252.34 ⇒ https://jenkins.nerdwallet.io/github-webhook/ with timestamp Tue Aug 29 10:44:40 PDT 2017 processed in 0.44 sec
    Checking branch github-reading

  Getting remote pull request #39...

  1 branches were processed

  Checking pull-requests...

    Checking pull request #39
      ‘Jenkinsfile’ found
ERROR: Server returned HTTP response code: -1, message: 'null' for URL: https://api.github.com/repos/NerdWallet/museum/pulls/39
java.net.SocketException: Socket closed

I will spend some time on this today or tomorrow.

[jglick]

@borgstrom any progress?

[ErikDahlinghaus]

This is really causing problems at my organization. Any updates on this issue? Thanks.

[bitwiseman]

@borgstrom Progress?
@ErikDahlinghaus Are you still seeing the issue? Perhaps this is a non-repro at this point.

[borgstrom]

We have not had any problems in a long time. I consider this abandoned at this point.

[ghost]

Could this pull request be reopened? The current retry mechanism does not have a delay and it fails when there are intermittent network issues.

[bitwiseman]

@bogdansuta-intel
It seems the original PR owner isn't seeing this issue any more. I encourage you to take these commits, make any further changes you deem needed and submit a new PR.