[FEATURE] Eager Refresh support when using SkipMemoryCache (L2-only scenarios)

[Coelho04]

Problem

We wanted to use SkipMemoryCache = true to avoid memory pressure on our pods, while still benefiting from Eager Refresh to keep our Redis cache warm and reduce latency.
However, after enabling SkipMemoryCache = true, we noticed Eager Refresh completely stopped firing — zero EagerRefresh events in our OpenTelemetry metrics.

Solution

I believe that the solution would be when validating that the distributed entry is valid to also perform the check if it should execute the eager refresh.

Thank you for the amazing work.

[Coelho04]

I believe the solution is something like this

if (distributedEntryIsValid)
						{
							isStale = false;
							entry = FusionCacheMemoryEntry<TValue>.CreateFromOtherEntry(distributedEntry!, options);
							
							if (isRealFactory && entry.ShouldEagerlyRefresh())
							{
								// RESET EAGER REFRESH
								if (entry!.Metadata is not null)
									entry.Metadata.EagerExpirationTimestamp = null;

								if (_logger?.IsEnabled(LogLevel.Trace) ?? false)
									_logger.Log(LogLevel.Trace, "FUSION [N={CacheName} I={CacheInstanceId}] (O={CacheOperationId} K={CacheKey}): should eagerly refresh", CacheName, InstanceId, operationId, key);

								// TRY TO GET THE MEMORY LOCK WITHOUT WAITING, SO THAT ONLY THE FIRST ONE WILL ACTUALLY REFRESH THE ENTRY
								var tmpMemoryLockObj = await AcquireMemoryLockAsync(operationId, key, TimeSpan.Zero, token).ConfigureAwait(false);
								if (tmpMemoryLockObj is null)
								{
									if (_logger?.IsEnabled(LogLevel.Trace) ?? false)
										_logger.Log(LogLevel.Trace, "FUSION [N={CacheName} I={CacheInstanceId}] (O={CacheOperationId} K={CacheKey}): eager refresh already occurring", CacheName, InstanceId, operationId, key);
								}
								else
								{
									_ = Task.Run(async () =>
									{
										await MaybeExecuteEagerRefreshWithAsyncFactoryAsync<TValue>(operationId, key, originalKey, tags, factory, options, entry!, tmpMemoryLockObj, token).ConfigureAwait(false);
									});
								}
							}
						}

Adding this code in both checks if (distributedEntryIsValid) would maybe work