Hi @stefanprodan ,
When deploying a lots of api services(use distributed cache to store counter) behind load balance, every instance need to get conuter and use process' lock to increse count, as the code below
using (await AsyncLock.WriterLockAsync(counterId).ConfigureAwait(false))
{
var entry = await _counterStore.GetAsync(counterId, cancellationToken);
if (entry.HasValue)
{
// entry has not expired
if (entry.Value.Timestamp + rule.PeriodTimespan.Value >= DateTime.UtcNow)
{
// increment request count
var totalCount = entry.Value.Count + _config.RateIncrementer?.Invoke() ?? 1;
// deep copy
counter = new RateLimitCounter
{
Timestamp = entry.Value.Timestamp,
Count = totalCount
};
}
}
// stores: id (string) - timestamp (datetime) - total_requests (long)
await _counterStore.SetAsync(counterId, counter, rule.PeriodTimespan.Value, cancellationToken);
}
I think it has concurrency issues, counter will not work correctly when a lots of requests come in. Every instance's counter count by themself then save to the cache store, the last one will cover the previous one.
Why not use Token Bucket algorithm to achieve this feature.
Hi @stefanprodan ,
When deploying a lots of api services(use distributed cache to store counter) behind load balance, every instance need to get conuter and use process' lock to increse count, as the code below
I think it has concurrency issues, counter will not work correctly when a lots of requests come in. Every instance's counter count by themself then save to the cache store, the last one will cover the previous one.
Why not use Token Bucket algorithm to achieve this feature.