How do I add external Redis support to an existing Docker install?

[ajtatum]

I have a Managed Redis Instance hosted on Digital Ocean. Their managed instances are running Redis 6.

username = default
password = [my password]
host = [my server].db.ondigitalocean.com
port = [my port]

The connection string looks like:

rediss://default:[my password]@[my server].db.ondigitalocean.com:[my port]

Looking at the documentation, it looks like you have you can only specify the Redis Server via TCP (which is fine I suppose), but I don't see how I supply the username and password for that Redis server. Additionally, it would be nice if I could specify which database to use so that it's not sharing the same database as my other applications.

Is this currently supported? If so, how can I make the change to my existing Docker image environment variables?

If it's not supported, let me know if I should create a feature request or if this can be converted to one.

Thanks,
AJ

[acelaya]

I'll check later today how it's done and come back to you.

If it's not possible, I'll convert this in a feature request.

Just in case, have you tried if it works with the redis://... URI?

[ajtatum]

I did not. How would I go about updating the configuration to try it out?

[acelaya]

It depends. How are you running Shlink? Docker, self hosted swoole or nginx/apache?

[ajtatum]

I went with Docker, but I'm starting to double think that decision as I'm not too familiar with actually running Docker in production. It was the easiest to get up and running though. I have nginx running as a proxy server in front of it, as suggested in the documentation.

[acelaya]

Cool.

You can use whatever option suits your needs. I was asking because setting new config options is different.

For the docker image you can just pass the REDIS_SERVERS env var with the URI.

https://shlink.io/documentation/install-docker-image/#use-an-external-redis-server

However, I have to admit redis is quite underused by Shlink at the moment. You won't get very high benefits from it, unless you plan to run a cluster of Shlink instances.

[ajtatum]

Cool, well I'll try it out. I don't plan to run a cluster of Shlink instances. I was hoping it would cache Urls and whatnot to save it from hitting the database every time someone clicks a link. Are there upcoming plans to use it more?

[acelaya]

Shlink used to cache more stuff, but it didn't add a significant performance improvement, so I ended up removing it for the sake of simplifying the code.

I have done some load tests in the past, and in general, performance is good.

The biggest bottleneck is in loading big amounts of visits from the API (I have documented some of that here #780).

However, that does not affect end users.

So there's no specific plans on caching stuff just for the sake of it, but I will if there's a clear need under high load scenarios. And when that happens, redis will be one of the supported systems for caching.