private ScaleStatus GetScaleStatusCore()

in src/WebJobs.Extensions.CosmosDB/Trigger/CosmosDBScaleMonitor.cs [64:155]
73 lines of code
13 McCabe index (conditional complexity)

        private ScaleStatus GetScaleStatusCore(int workerCount, CosmosDBTriggerMetrics[] metrics)
        {
            ScaleStatus status = new ScaleStatus
            {
                Vote = ScaleVote.None
            };

            const int NumberOfSamplesToConsider = 5;

            // Unable to determine the correct vote with no metrics.
            if (metrics == null)
            {
                return status;
            }

            // We shouldn't assign more workers than there are partitions (Cosmos DB, Event Hub, Service Bus Queue/Topic)
            // This check is first, because it is independent of load or number of samples.
            int partitionCount = metrics.Length > 0 ? metrics.Last().PartitionCount : 0;
            if (partitionCount > 0 && partitionCount < workerCount)
            {
                status.Vote = ScaleVote.ScaleIn;
                _logger.LogInformation(Events.OnScaling, string.Format($"WorkerCount ({workerCount}) > PartitionCount ({partitionCount})."));
                _logger.LogInformation(Events.OnScaling, string.Format($"Number of instances ({workerCount}) is too high relative to number " +
                                                     $"of partitions for container ({_monitoredContainer.Id}, {partitionCount})."));
                return status;
            }

            // At least 5 samples are required to make a scale decision for the rest of the checks.
            if (metrics.Length < NumberOfSamplesToConsider)
            {
                return status;
            }

            // Maintain a minimum ratio of 1 worker per 1,000 items of remaining work.
            long latestRemainingWork = metrics.Last().RemainingWork;
            if (latestRemainingWork > workerCount * 1000)
            {
                status.Vote = ScaleVote.ScaleOut;
                _logger.LogInformation(Events.OnScaling, string.Format($"RemainingWork ({latestRemainingWork}) > WorkerCount ({workerCount}) * 1,000."));
                _logger.LogInformation(Events.OnScaling, string.Format($"Remaining work for container ({_monitoredContainer.Id}, {latestRemainingWork}) " +
                                                     $"is too high relative to the number of instances ({workerCount})."));
                return status;
            }

            bool documentsWaiting = metrics.All(m => m.RemainingWork > 0);
            if (documentsWaiting && partitionCount > 0 && partitionCount > workerCount)
            {
                status.Vote = ScaleVote.ScaleOut;
                _logger.LogInformation(Events.OnScaling, string.Format($"CosmosDB container '{_monitoredContainer.Id}' has documents waiting to be processed."));
                _logger.LogInformation(Events.OnScaling, string.Format($"There are {workerCount} instances relative to {partitionCount} partitions."));
                return status;
            }

            // Check to see if the trigger source has been empty for a while. Only if all trigger sources are empty do we scale down.
            bool isIdle = metrics.All(m => m.RemainingWork == 0);
            if (isIdle)
            {
                status.Vote = ScaleVote.ScaleIn;
                _logger.LogInformation(Events.OnScaling, string.Format($"'{_monitoredContainer.Id}' is idle."));
                return status;
            }

            // Samples are in chronological order. Check for a continuous increase in work remaining.
            // If detected, this results in an automatic scale out for the site container.
            bool remainingWorkIncreasing =
                IsTrueForLast(
                    metrics,
                    NumberOfSamplesToConsider,
                    (prev, next) => prev.RemainingWork < next.RemainingWork) && metrics[0].RemainingWork > 0;
            if (remainingWorkIncreasing)
            {
                status.Vote = ScaleVote.ScaleOut;
                _logger.LogInformation(Events.OnScaling, $"Remaining work is increasing for '{_monitoredContainer.Id}'.");
                return status;
            }

            bool remainingWorkDecreasing =
                IsTrueForLast(
                    metrics,
                    NumberOfSamplesToConsider,
                    (prev, next) => prev.RemainingWork > next.RemainingWork);
            if (remainingWorkDecreasing)
            {
                status.Vote = ScaleVote.ScaleIn;
                _logger.LogInformation(Events.OnScaling, $"Remaining work is decreasing for '{_monitoredContainer.Id}'.");
                return status;
            }

            _logger.LogInformation(Events.OnScaling, $"CosmosDB container '{_monitoredContainer.Id}' is steady.");

            return status;
        }