Do not crash/error if Redis is unreachable for request limits

[arkocal]

It is possible to achieve High Availability in Redis by master-slave replication, but most likely the costs of operating replicas is not worth for the short unexpected downtimes. I thinks it is better to temporarily disable rate limits by skipping the queries when Redis is not reachable. This way, we can offer HA in frontend, even when we don't in Redis with minimal changes in operation.

[oguzcankirmemis]

You mean queries will be still done but if frontend can't reach to redis it ignores the query?

[arkocal]

Yes, precisely.

[oguzcankirmemis]

I should also verify the redis code, that it works the way we expect, I will take a look at this while I do that

[wagmarcel]

But redis is also used by mqtt gateway to store Tokens and will be used in future by wss.

[arkocal]

From a scalibility perspective, it is better to use a seperate redis instance for that purpose.

See the first bullet list in Chapter 10.2 in Redis in Action: https://redislabs.com/ebook/part-3-next-steps/chapter-10-scaling-redis/10-2-scaling-writes-and-memory-capacity/

[arkocal]

In case seperating the instances is not necessary, they can also use a single redis with HA, but I think it is better to have the option not to depend on Redis HA from the front end, it would not cause any problems when redis is also replicated, which would be useful/necessary for wss.

[oguzcankirmemis]

#202 should be able to skip redis if it is not reachable, meaning it will allow the request.