Error: Connection timeout

[mspoerer]

Hi everyone,
we are currently experiencing connection timeouts in our Azure pipelines. During the deployment process, we are observing strange behavior with the Redis client. After receiving an "Error: Connection timeout," it appears that multiple clients attempt to reconnect themselves.

Upon investigation, we found the following lines of code, which do not make sense from our perspective, since the Redis client should handle reconnections itself via the reconnectStrategy.

https://github.com/trieb-work/nextjs-turbo-redis-cache/blob/7622dd5f9cc681353177776b284a98bb0b136078/src/RedisStringsHandler.ts#L193C11-L202C13

Are we missing something here, or are the mentioned lines obsolete?

[tilman]

Hey, I have added this reconnect line since in our case (Google cloud deployment) the redis client was not reconnecting automatically after receiving an connection lost error in normal operation in production. After adding this explicitly it was working again.

Next.js is using a multi worker approach (especially during build) to build multiple pages in parallel. Each worker, will start it's own instance of cache handler, resulting in multiple clients connecting to redis. Are you sure that your observed behaviour is from this line of code or rather from Next.js worker architecture?

From my understanding of the redis package code and testing the function, this.client.connect did not have any effect if the client was already connected.