记录异常并重新抛出的 Polly 策略

Question

I consider to use Polly to create policy to log exception and rethrow. I didn't find an existing method that allow it out of the box, but some options that I see are

Fallback

// Specify a substitute value or func, calling an action (e.g. for logging)
// if the fallback is invoked.
Policy.Handle<Whatever>()
.Fallback<UserAvatar>(UserAvatar.Blank,
    onFallback: (exception, context) =>
    {
        _logger.Log(exception, context);
        throw exception;
    });

Question: is it ok to throw exception from Fallback?

Timeout

Policy.Timeout(1, T30meoutStrategy.Pessimistic,
(context, timespan, task) =>
{
    // ContinueWith important!: the abandoned task may very well still be executing,
    // when the caller times out on waiting for it!
    task.ContinueWith(t =>
    {
        if (t.IsFaulted)
        {
            logger.Error(context,t.Exception);
            throw exception;
        }
    });
}

Or Retry

Policy.Handle<DivideByZeroException>().Retry(0,
(exception, retryCount) =>
{
    logger.Error(context,exception);
    throw exception;
});

Question: is 0 retries supported?

Or KISS and write simple try/catch with throw by myself.

Which of these methods is better? What are your recommendation?

Answer 1

If you do not already have Polly in the mix, try/catch would seem simplest.

If you already have Polly in the mix, FallbackPolicy can safely be re-purposed in the way you suggest. The onFallback delegate and fallback action or value are not governed by the .Handle<>() clauses of the Policy , so you can safely rethrow an exception from within the onFallback delegate.

Policy<UserAvatar>.Handle<Whatever>()
.Fallback<UserAvatar>(UserAvatar.Blank,
    onFallback: (exception, context) =>
    {
        _logger.Log(exception, context);
        throw exception;
    });

---

The approach your question outlines with TimeoutPolicy would only capture exceptions thrown by delegates the caller had earlier walked away from due to timeout, and only in TimeoutMode.Pessimistic ; not all exceptions.

---

The approach your question outlines with .Retry(0, ...) would not work. If no retries are specified, the onRetry delegate would not be invoked.

---

To avoid the untidiness of repurposing FallbackPolicy , you could also code your own LogThenRethrowPolicy , within Polly's structures. This commit (which added the simple NoOpPolicy ) exemplifies the minimum necessary to add a new policy. You could add an implementation similar to NoOpPolicy but just try { } catch { /* log; rethrow */ } try { } catch { /* log; rethrow */ }

---

EDIT January 2019 : Polly.Contrib now also contains a Polly.Contrib.LoggingPolicy which can help with this.

Answer 2

https://github.com/App-vNext/Polly-Samples/blob/master/PollyDemos/Async/AsyncDemo02_WaitAndRetryNTimes.cs shows that you can use the onRetry: option, at least for WaitAndRetryAsync. I haven't looked at the others yet.

HttpPolicyExtensions
.HandleTransientHttpError()
.WaitAndRetryAsync(3,
    retryAttempt => TimeSpan.FromSeconds(Math.Pow(2, retryAttempt))  // exponential back-off: 2, 4, 8 etc
                    + TimeSpan.FromMilliseconds(Jitterer.Next(0, 1000)), // plus some jitter: up to 1 second
    onRetry: (response, calculatedWaitDuration) =>
    {
        logger.LogError($"Failed attempt. Waited for {calculatedWaitDuration}. Retrying. {response.Exception.Message} - {response.Exception.StackTrace}");
    }
);

Answer 3

Here my solution with generic method

public async Task<T> PollyRetry<T>(
        Func<Task<T>> action)
    {

        bool hasFallback = false;
        Exception ex = null;

        var fallbackPolicy = Policy<T>.Handle<Exception>().FallbackAsync(
            default(T), d =>
            {
                //log final exception

                ex = d.Exception;

                hasFallback = true;
                return Task.FromResult(new { });

            });

        var retryPolicy = Policy
            .Handle<Exception>()
            .WaitAndRetryAsync(3,
                retryAttempt => TimeSpan.FromSeconds(Math.Pow(2, retryAttempt)),
                (res, timeSpan, context) =>
                {
                    //log exception
                });

        var policyResult = await fallbackPolicy.WrapAsync(retryPolicy).ExecuteAndCaptureAsync(action);

        if (hasFallback && ex != null)
            throw ex;

        return policyResult.Result;
    }

---

//call service with retry logic
        TestResponse response = await _pollyRetryService.PollyRetry(async () =>
        {
            return await _testService.Test(input);

        });