client is closed error

[fco-fbatch]

Getting client is closed error while trying to use the latest version.

@erhant you can check the code I am getting an error here. And my redisClient and hollowdb creation process is here.

The error I am getting

/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/@redis/client/dist/lib/client/index.js:490
        return Promise.reject(new errors_1.ClientClosedError());
                              ^

ClientClosedError: The client is closed
    at RedisClient._RedisClient_sendCommand (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/@redis/client/dist/lib/client/index.js:490:31)
    at RedisClient.commandsExecutor (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/@redis/client/dist/lib/client/index.js:188:154)
    at BaseClass.<computed> [as SET] (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/@redis/client/dist/lib/commander.js:8:29)
    at RedisCache.put (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts-redis/lib/cjs/redisCache.js:277:31)
    at CacheableStateEvaluator.putInCache (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts/lib/cjs/core/modules/impl/CacheableStateEvaluator.js:126:26)
    at CacheableStateEvaluator.onStateEvaluated (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts/lib/cjs/core/modules/impl/CacheableStateEvaluator.js:67:20)
    at CacheableStateEvaluator.doReadState (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts/lib/cjs/core/modules/impl/DefaultStateEvaluator.js:244:24)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async CacheableStateEvaluator.eval (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts/lib/cjs/core/modules/impl/CacheableStateEvaluator.js:62:16)
    at async HandlerBasedContract.readState (/home/fco/Documents/GitHub/hollowdb-crud-microservices/get-service/node_modules/warp-contracts/lib/cjs/contract/HandlerBasedContract.js:121:28)

Node.js v18.12.1
error Command failed with exit code 1.

[fco-fbatch]

Bonus explenation,

If I try to manually call connect() before the hollowdb.readState(), either by directly using redisClient or using the redisClient by underlying storage() method, I am getting the connection is already opened error.

If I omit using connect() I get The client is closed error.

[erhant]

Apparently, just using useKVStorageFactory alone does not produce this error, but using multiple cache options (useStateCache and useContractCache) do cause it.

[fco-fbatch]

As a solution, I am offering adding a separate opt isManaged when creating the RedisCache. If isManaged is true then the underlying warp instance cannot close or open connections to redis only the caller can. You can check this branch for the proposed changes. I also changed the redisClientType.

[erhant]

#8 introduces this temporary fix, along with major changes to the code. now, the client is created within the cache as is the case in LMDB.

[erhant]

@fco-fbatch the recent pull seems to fix the error, without the workaround!
still not closing this as we need further testing

[erhant]

#9 now allows client to be added from outside, which internally makes isManaged = true and expects the clilent to make open and close calls.

[erhant]

probably should be mentioned, the other SortKeyCache implementations are targeted for local environments (as per the lazy-evaluation on client-side requirements) however Redis cache aims to provide a single online cache for multiple clients.

This may be the root cause of open and close bugs.