Fix possible endless loop while polling curl socket

[CurtizJ]

In case of high frequency of received signals (more than once per second) the field now will never be updated and it can lead to exceeding of timeout or endless loop if socket will never be ready. For instance in ClickHouse signals SIGUSR1 and SIGUSR2 are being sent each second by internal query profiler and we faced this issue.

Pull Request Checklist

Please leverage this checklist as a reminder to address commonly occurring feedback when submitting a pull request to make sure your PR can be reviewed quickly:

See the detailed list in the contributing guide.

• C++ Guidelines
• Doxygen docs
• Unit tests
• No unwanted commits/changes
• Descriptive title/description
  • PR is single purpose
  • Related issue listed
• Comments in source
• No typos
• Update changelog
• Not work-in-progress
• External references or docs updated
• Self review of PR done
• Any breaking changes?

[github-actions]

Thank you for your contribution @CurtizJ! We will review the pull request and get back to you soon.

[LarryOsterman]

Oops. Extremely good catch.

Is it possible to write a test for this?

[CurtizJ]

Yes, I'll try to add a test.

[ahsonkhan]

That's a great catch! I am curious, how did you discover this? Did you actually hit the endless loop, in your use case?

[CurtizJ]

how did you discover this?

We use this sdk for integration with azure blob storage in ClickHouse. I ran a stress test with heavy queries and after finishing it I noticed some hung queries with stacktrace on the line changed in this PR. I don't know why poll hasn't received any events, maybe because the connection was by timeout, but I didn't investigate it. After fixing this the issue has disappeared.

[CurtizJ]

Sorry for the delay, it appeared to be harder to add test than I thought.

[ahsonkhan]

@LarryOsterman given this is a great reliability bug fix, I propose we merge it.

I'll make a changelog update PR, for core, attributing @CurtizJ for fixing this issue. Thanks!