What is the Redis rate limit at various plan sizes? I am receiving Too many requests. Please try again later. errors, and can’t find the rate limit information anywhere online.
Note that I have ~100 active connections on a Standard Redis plan (1,000 active connections allowed), so this does not appear to be due to that cause.
What can I do to increase that rate limit?
I use Redis for background job queues/management; these workloads are necessarily bursty. I can’t imagine this is an uncommon use case, and I’m frankly pretty surprised to be seeing these errors given my relatively low concurrency, connection utilization compared to plan size, and standard tooling (it’s all BullMQ — nothing custom and no additional redis operations outside of BullMQs standard stuff).
The service info page for a Redis service on the Render Dashboard shows the connection limit for your particular instance type. Click the service and it’s in the second box of info labeled “Redis Instance”. As you stated, a Standard type Redis instance has a 1,000 connection limit.
Connection limits are pinned to the plan type, both the instance type picker and our pricing page ( Pricing | Render ) shows connection limits per type. The next step up, Pro, offers 5,000 connections.
Aside from upgrading the instance type, you can look into Redis connection pooling for your services’ language/framework/library (or libraries) in order to cut overall connection usage down.
If you feel that you are not reaching the 1,000 connection limit, contact us directly for further assistance.
Sorry Jason, I think you misread my message. This is not the problem. I am an order of magnitude below the connection limit. I am getting rate limited, not bumping up against the connection cap. Please see my original message, and let me know the best path forward.
I’ve now gotten an answer from internal support. Regardless of total connection cap, Render has a soft cap of 5 connections per second from external sources (eg, a server on AWS), likely causing my issue. Posting this here in case others run into the same issue.
For future folks who run into this issue, my solution is to bite the bullet and move Redis over to a major cloud provider — I’ve implemented connection pooling, etc, but at any reasonable background job concurrency, I’m going to smash that 5 connections / second in bursts.
