Add multitenant support to allow keeping tenant specific data separated for the cached models. #81
Conversation
$eagerLoad = $this->eagerLoad ?? []; $model = $this->model ?? $this; $query = $this->query ?? app(Builder::class); with $eagerLoad = $this->retrieveEagerLoad(); $model = $this->retrieveCacheModel(); $query = $this->retrieveCacheQuery(); to avoid name collisions in case the model contains one of the 'eagerLoad', 'mode' or 'query' attributes - added makeCachePrefix and used it to prefix cache tags and keyDifferentiator and class name tag with the value returned by the 'getCachePrefix' method of the model This method can be used in a multitenant implementation where an object with the same key and tags is cached for each tenant database connection. - added CacheGlobal class to be a placeholder for the global static cache disable flag. The cached value using $cache might generate unpredictable behaviour in case multiple requests are made to the server in the same time.
|
I would recommend creating individual redis connections for each tenant, and handle it via configuration, instead of implementing it in the code. I might consider addressing this when I refactor the way keys are generated. Keep working on this by all means, and I will take it into consideration when adding that functionality. |
|
Sorry, didn't mean to close it. |
|
I've been reading through you proposed changes some more, and feel this PR contains too many different changes (this that are not necessarily related to just implementing cache prefixing, but other refactoring as well). Would you mind stripping this PR down to only cache prefixing without any of the other refactoring, if possible? I don't necessarily want to merge all those changes. |
|
Actually, sorry to flip-flop on this so much. I will be implementing cache key prefixes now, but not using this PR, as I don't want to incorporate everything it has. Thanks for submitting this, though. Sorry for not accepting your PR, hopefully next time! :) |
|
Hi @lucian-dragomir this should be fixed in version 0.3.31. Please let me know how that works for you. |
|
Hi @mikebronner, |
This is my proposal of adding multi-tenant support to the cached models by implementing at model level (in a base model class or using a trait) a new method getCachePrefix that can be used to differentiate cache tags and keys by prefixing them with the tenant context (in the provided example the method is returning the connection name concatenated with the tenant database name).