- What is your product version? ABP Nuget packages 6.6.1
- What is your product type (Angular or MVC)? ANGULAR
- What is product framework type (.net framework or .net core)? .NET 5
Yesterday we rolled out a new version of our application where we updated from using version 4.12 of the ABP packages to 6.6.1 We found an unexpected issue with our caching. We implemented our own hybrid solution, where we try to get values from the memory cache which has a small expiry and if it's not there, we get it from Redis and put in the Memory cache for a next time. Previously we had a small amount of redis cache hits and most of the work was done by memory cache. We made some alterations to make it possible to upgrade from 4.12 to 6.6.1, but unfortunately the colleague that made these changes for the caching has since left the company.
Can you help us understand what might have changed in ABP that could cause this, or how our solution might be responsible for this. If you need any further info, please let me know!
`namespace xxx.xxx.Web.Caching { using System; using System.Collections.Generic; using System.Threading.Tasks; using Abp.Runtime.Caching; using Abp.Runtime.Caching.Memory; using Abp.Runtime.Caching.Redis; using Abp.Timing;
/// <summary>
/// Create a cache existing of two caching layers, the first being a memory cache, the second being a Redis cache.
/// The memory cache expires quickly, so a change on another AppService instance is -through the use of the Redis cache- propagated quickly to the memory cache of this instance
/// Create, update and delete actions are always applied to the memory cache AND the Redis cache
/// Getting an item from cache will try to get it from the memory cache. If this doesn't succeed, it will try to get it from the Redis cache.
/// </summary>
public class MemoryAndRedisCache : CacheBase
{
private readonly AbpMemoryCache _memoryCache;
private readonly AbpRedisCache _redisCache;
/// <summary>
/// Constructor.
/// </summary>
public MemoryAndRedisCache(string name, AbpMemoryCacheManager memoryCacheManager, AbpRedisCacheManager redisCacheManager) : base(name)
{
// Set expiry to 1 minute for the memory cache, so a change on another AppService instance is propagated quickly to the memory cache of this instance
_memoryCache = (AbpMemoryCache)memoryCacheManager.GetCache(name + "_memory");
_memoryCache.DefaultAbsoluteExpireTime = Clock.Now.AddMinutes(1);
// Expiry of the Redis cache is done through normal configuration in ABP
_redisCache = (AbpRedisCache)redisCacheManager.GetCache(name);
}
public override bool TryGetValue(string key, out object value)
{
// Try to get the value from memory
var memorySuccess = _memoryCache.TryGetValue(key, out value);
if (memorySuccess)
{
var expiryInfo = _memoryCache.DefaultAbsoluteExpireTime != null ? _memoryCache.DefaultAbsoluteExpireTime.ToString() : _memoryCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [memory] [{expiryInfo}]");
return true;
}
// Getting from memory failed, try to get it from Redis
var redisSuccess = _redisCache.TryGetValue(key, out value);
if (redisSuccess)
{
var expiryInfo = _redisCache.DefaultAbsoluteExpireTime != null ? _redisCache.DefaultAbsoluteExpireTime.ToString() : _redisCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [redis] [{expiryInfo}]");
// We got the value from Redis, now store it in memory cache for repeated use
_memoryCache.Set(key, value, null, null);
return true;
}
return false;
}
public override object GetOrDefault(string key)
{
// Try to get the value from memory
var value = _memoryCache.GetOrDefault(key);
if (value != null)
{
var expiryInfo = _memoryCache.DefaultAbsoluteExpireTime != null ? _memoryCache.DefaultAbsoluteExpireTime.ToString() : _memoryCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [memory] [{expiryInfo}]");
return value;
}
// Getting from memory failed, try to get it from Redis
value = _redisCache.GetOrDefault(key);
if (value != null)
{
var expiryInfo = _redisCache.DefaultAbsoluteExpireTime != null ? _redisCache.DefaultAbsoluteExpireTime.ToString() : _redisCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [redis] [{expiryInfo}]");
// We got the value from Redis, now store it in memory cache for repeated use
_memoryCache.Set(key, value, null, null);
}
else
{
Logger.Debug($"Get from {this.Name}: {key} [source]");
// The caching logic of ABP will call the function to get the data from the source - no need to do anything
}
return value;
}
public override object[] GetOrDefault(string[] keys)
{
var values = _memoryCache.GetOrDefault(keys);
for (var t = 0; t < keys.Length; t++)
{
if (values[t] == null)
{
values[t] = _redisCache.GetOrDefault(keys[t]);
}
}
return values;
}
public override async Task<object> GetOrDefaultAsync(string key)
{
// Try to get the value from memory
var value = await _memoryCache.GetOrDefaultAsync(key);
if (value != null)
{
var expiryInfo = _memoryCache.DefaultAbsoluteExpireTime != null ? _memoryCache.DefaultAbsoluteExpireTime.ToString() : _memoryCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [memory] [{expiryInfo}]");
return value;
}
// Getting from memory failed, try to get it from Redis
value = await _redisCache.GetOrDefaultAsync(key);
if (value != null)
{
var expiryInfo = _redisCache.DefaultAbsoluteExpireTime != null ? _redisCache.DefaultAbsoluteExpireTime.ToString() : _redisCache.DefaultSlidingExpireTime.ToString();
Logger.Debug($"Get from {this.Name}: {key} [redis] [{expiryInfo}]");
// We got the value from Redis, now store it in memory cache for repeated use
await _memoryCache.SetAsync(key, value, null, null);
}
else
{
Logger.Debug($"Get from {this.Name}: {key} [source]");
// The caching logic of ABP will call the function to get the data from the source - no need to do anything
}
return value;
}
public override async Task<object[]> GetOrDefaultAsync(string[] keys)
{
var values = await _memoryCache.GetOrDefaultAsync(keys);
for (var t = 0; t < keys.Length; t++)
{
if (values[t] == null)
{
values[t] = await _redisCache.GetOrDefaultAsync(keys[t]);
}
}
return values;
}
public override void Set(string key, object value, TimeSpan? slidingExpireTime = null, DateTimeOffset? absoluteExpireTime = null)
{
_memoryCache.Set(key, value, null, null);
_redisCache.Set(key, value, null, null);
}
public override async Task SetAsync(string key, object value, TimeSpan? slidingExpireTime = null, DateTimeOffset? absoluteExpireTime = null)
{
await _memoryCache.SetAsync(key, value, null, null);
await _redisCache.SetAsync(key, value, null, null);
}
public override void Set(KeyValuePair<string, object>[] pairs, TimeSpan? slidingExpireTime = null, DateTimeOffset? absoluteExpireTime = null)
{
_memoryCache.Set(pairs, null, null);
_redisCache.Set(pairs, null, null);
}
public override async Task SetAsync(KeyValuePair<string, object>[] pairs, TimeSpan? slidingExpireTime = null, DateTimeOffset? absoluteExpireTime = null)
{
await _memoryCache.SetAsync(pairs, null);
await _redisCache.SetAsync(pairs, null, null);
}
public override void Remove(string key)
{
_memoryCache.Remove(key);
_redisCache.Remove(key);
}
public override async Task RemoveAsync(string key)
{
await _memoryCache.RemoveAsync(key);
await _redisCache.RemoveAsync(key);
}
public override void Remove(string[] keys)
{
_memoryCache.Remove(keys);
_redisCache.Remove(keys);
}
public override async Task RemoveAsync(string[] keys)
{
await _memoryCache.RemoveAsync(keys);
await _redisCache.RemoveAsync(keys);
}
public override void Clear()
{
_memoryCache.Clear();
_redisCache.Clear();
}
public override void Dispose()
{
_memoryCache.Dispose();
_redisCache.Dispose();
}
}
} `
3 Answer(s)
-
0
Hi @web2worknl Abp already has that implementation. It is called
PerRequestRedisCache
. We may offer you to use it instead.See: https://aspnetboilerplate.com/Pages/Documents/Caching#per-request-redis-cache https://aspnetboilerplate.com/Pages/Documents/PerRequestRedisCache
-
0
Thanks for the suggestion! I was unaware that ABP had something similar out of the box 👍.
I have a few questions about the implementation of the PerRequestRedisCache though
- Does this cache stuff to memory like the MemoryCache implementation does, or does it cache for the specific current connection so THIS user doesn't have to go to redis if they need the same value again
- We have a scale out solution with multiple instances, so we've set the expiry on the memory cache really low so it expires quickly and is updated from Redis frequently. Is it possible to set the expiry of the memorycache with PerRequestRedisCache?
-
0
Hi @web2worknl
It caches items requested from redis per request. In every request it will be cached locally. When the cache changed in the current instance of your project, local cached instance of the cache will also be changed. But if you change it in another instance of your application current instance will not be able to get in current httprequest.