fix: Adjust big segments logic to prevent duplicate polls. #225
Conversation
| @@ -43,8 +44,9 @@ Logger logger | |||
| _taskExecutor = taskExecutor; | |||
| _logger = logger; | |||
|
|
|||
| _initialPoll = Task.Run(PollStoreAndUpdateStatusAsync); | |||
There was a problem hiding this comment.
Manually run the first task instead of using the repeating task.
| return ret.Value; | ||
| } | ||
| return AsyncUtils.WaitSafely(() => PollStoreAndUpdateStatusAsync()); | ||
| return ret ?? _initialPoll.GetAwaiter().GetResult(); |
There was a problem hiding this comment.
If there is no value, then that first poll has not completed. Wait for it to complete.
There was a problem hiding this comment.
Is it possible that the initial task is still executing by the time the first repeating task runs? I suppose it may be extremely unlikely given the real world values of StatusPollInterval.
There was a problem hiding this comment.
It theoretically is possible. I did consider if there was a way to use the most recent task, but the complexity didn't seem worth it. If at any point a repeating task does complete, then we would ideally hit the first code path subsequently.
| return ret.Value; | ||
| } | ||
| return AsyncUtils.WaitSafely(() => PollStoreAndUpdateStatusAsync()); | ||
| return ret ?? _initialPoll.GetAwaiter().GetResult(); |
There was a problem hiding this comment.
Is it possible that the initial task is still executing by the time the first repeating task runs? I suppose it may be extremely unlikely given the real world values of StatusPollInterval.
🤖 I have created a release *beep* *boop* --- ## [8.5.1](8.5.0...8.5.1) (2024-06-03) ### Bug Fixes * Adjust big segments logic to prevent duplicate polls. ([#225](#225)) ([4202165](4202165)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please). Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
The contract tests attempt to determine that the SDK is not making redundant polls, and sometimes this test would fail.
Theoretically this could also affect a server handling requests on many threads which concurrently request status before the first request has completed.