Implement rate-limit for outgoing requests

[adeelx]

Problem

I'm using 1.1.1.1 as my upstream as follows:

upstreams:
  strategy: parallel_best
  init:
    strategy: fast
  groups:
    default:
    - tcp-tls:1.1.1.1:853 # cloudflare
    - tcp-tls:1.0.0.1:853 # cloudflare

Whenever I restart the blocky container I run afoul of CloudFlare's 10QPS limit because all the devices on the network are simultaneously requesting IPs. To counter this, I have had to put blocky behind dnsmasq and on every restart I have to first switch dnsmasq to use Google DNS, then start blocky and finally once blocky shows that it can handle queries I switch dnsmasq to use blocky.

Do note that this happens only when the blocky DNS server has been down for ~30mins or so and the network doesn't have any DNS resolvers which causes devices to retry queries and hence builds up the query volume.

Relevant Links

• Is there any rate-limiting for 1.1.1.1 DNS queries?
• The rate limit isn't static and 10qps is pretty rare.
• Throttling or other issue

The Ask

It will be helpful if blocky can be configured to not exceed a given QPS limit on a global level. Something along the lines of follows:

upstreams:
  qps_limit: 10 # or another sane default
  strategy: parallel_best
  init:
    strategy: fast
  groups:
    default:
    - tcp-tls:1.1.1.1:853 # cloudflare
    - tcp-tls:1.0.0.1:853 # cloudflare

This would make cold-starts of blocky much easier.

[ThinkChaos]

Does configuring the bootstrap DNS server to something else workaround the issue?

[adeelx]

Yes, configuring it to Google DNS alleviates the problem.

[github-actions]

This issue is stale because it has been open 90 days with no activity. Remove stale label or comment or this will be closed in 5 days.

[github-actions]

This issue was closed because it has been stalled for 5 days with no activity.