This is a Redis based ICacheProvider for NHibernate written in C# using ServiceStack.Redis.
- You can install using NuGet:
PM> Install-Package NHibernate.Caches.Redis
- Or build/install from source:
msbuild .\build\build.proj
and then look inside thebin
directory.
Configure NHibernate to use the custom cache provider:
<property name="cache.use_second_level_cache">true</property>
<property name="cache.use_query_cache">true</property>
<property name="cache.provider_class">NHibernate.Caches.Redis.RedisCacheProvider,
NHibernate.Caches.Redis</property>
Set the ConnectionMultiplexer
on the RedisCacheProvider
before creating your ISessionFactory
:
// Or use your IoC container to wire this up.
var connectionMultiplexer = ConnectionMultiplexer.Connect("localhost:6379");
RedisCacheProvider.SetConnectionMultiplexer(connectionMultiplexer);
using (var sessionFactory = ...)
{
// ...
}
// When your application exits:
connectionMultiplexer.Dispose();
Check out the NHibernate.Caches.Redis.Sample
project to learn more.
You can customize certain behavior with the RedisCacheProvider.SetOptions(options)
method. For example, you can control how objects are serialized into Redis.
Here is a JSON.NET ICacheSerializer
implementation. Once added to your project, you can then configure the options:
var options = new RedisCacheProviderOptions()
{
Serializer = new NhJsonCacheSerializer()
};
RedisCacheProvider.SetOptions(options);
Similar to other NHibernate cache providers, inside of the app/web.config
, a
custom configuration section can be added to configure each cache region. For
example, this feature allows you to control the expiration for a specific class
that you cache.
<configSections>
<section name="nhibernateRedisCache" type="NHibernate.Caches.Redis.RedisCacheProviderSection, NHibernate.Caches.Redis" />
</configSections>
<nhibernateRedisCache>
<caches>
<cache region="BlogPost" expiration="900" />
</caches>
</nhibernateRedisCache>
You may require that we gracefully continue to the database as if we "missed" the cache if an exception occurs. By default, this is what happens.
However, there is also the need of advanced exception handling scenarios. For example, imagine if you are using NHibernate in a web project and your Redis server is unavailable. You may not want NHibernate to continue to timeout for every NHibernate operation. You could do something similar to this:
public class RequestRecoveryRedisCache : RedisCache
{
public const string SkipNHibernateCacheKey = "__SkipNHibernateCache__";
public RequestRecoveryRedisCache(string regionName, IDictionary<string, string> properties, RedisCacheElement element, ConnectionMultiplexer connectionMultiplexer, RedisCacheProviderOptions options)
: base(regionName, properties, element, connectionMultiplexer, options)
{
}
public override object Get(object key)
{
if (HasFailedForThisHttpRequest()) return null;
return base.Get(key);
}
public override void Put(object key, object value)
{
if (HasFailedForThisHttpRequest()) return;
base.Put(key, value);
}
public override void Remove(object key)
{
if (HasFailedForThisHttpRequest()) return;
base.Remove(key);
}
public override void Clear()
{
if (HasFailedForThisHttpRequest()) return;
base.Clear();
}
public override void Destroy()
{
if (HasFailedForThisHttpRequest()) return;
base.Destroy();
}
public override void Lock(object key)
{
if (HasFailedForThisHttpRequest()) return;
base.Lock(key);
}
public override void Unlock(object key)
{
if (HasFailedForThisHttpRequest()) return;
base.Unlock(key);
}
private bool HasFailedForThisHttpRequest()
{
return HttpContext.Current.Items.Contains(SkipNHibernateCacheKey);
}
}
public class RequestRecoveryRedisCacheProvider : RedisCacheProvider
{
protected override RedisCache BuildCache(string regionName, IDictionary<string, string> properties, RedisCacheElement configElement, ConnectionMultiplexer connectionMultiplexer, RedisCacheProviderOptions options)
{
options.OnException = (e) =>
{
HttpContext.Current.Items[RequestRecoveryRedisCache.SkipNHibernateCacheKey] = true;
};
return new RequestRecoveryRedisCache(regionName, properties, configElement, connectionMultiplexer, options);
}
}
Then, use RequestRecoveryRedisCacheProvider
in your web.config
settings.
2.0.0
- Switch the Redis library to StackExchange.Redis because of licensing changes with ServiceStack.Redis. This obviously causes a few breaking changes with the constructors.
- Introduce
RedisCacheProvider.SetOptions
(so that, for example, you don't need to subclass to overrideOnException
). - Allow the serializer to be customized by implementing
ICacheSerializer
and setting theSerializer
on the options. The default serializer uses theNetDataContractSerializer
. - Customize which database the Redis connection uses with the
Database
option - The cache key no longer duplicates the region prefix. In previous
versions, caching an object with the type
MyApp.Models.Blog
and a region prefix ofv2
would use the keyv2:NHibernate-Cache:v2.ProcedureFlow.Core.Models.User:keys
. The key is nowv2:NHibernate-Cache:ProcedureFlow.Core.Models.User:keys
. - Allow the lock value to be customized. This is useful if you want to store information such as what machine/process generated the lock to help with debugging.
1.3.0
- Add the
OnException
method for sub-classing the cache client and handling exceptions.
1.2.1
- Update ServiceStack.Redis to 3.9.55.
1.2.0
- Allow the provider to gracefully continue when Redis is unavailable.
- Fix infinite loop when data in Redis was cleared.
1.1.0
- Added configuration section for customizing the cache regions.
- Added sample project.
1.0.0
- Initial release.
@MattiasJakobsson and @Tazer for helping switch over to StackExchange.Redis
.
Happy caching!