fix(routes): many requests for outdated routes by rate limiting #675
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #675 +/- ##
==========================================
+ Coverage 72.10% 72.26% +0.16%
==========================================
Files 32 32
Lines 2667 2668 +1
==========================================
+ Hits 1923 1928 +5
+ Misses 555 552 -3
+ Partials 189 188 -1 ☔ View full report in Codecov by Sentry. |
jooola
approved these changes
Jul 5, 2024
Fixes #673 In `routes.ListRoutes()` we have to find the matching server/node for every route in the network. We find the server by utilizing a cache that maps every private IP to the corresponding server. This cache has a feature that refreshes the list of servers if an entry can not be found. This is sensible, as the server might have been just created. This is also fatal, as this refresh happens for every single cache access. If there are a lot of routes in the network that do not belong to any server, we refresh the cache many times for each `ListRoutes()`. This is even more serious, as `ListRoutes()` is being called every 10-30 seconds (see #395). This commit introduces a `rate.Limiter` in the `AllServersCache` which only allows the refresh to happen every 30 seconds.
|
I rebased onto main to resolve the go.mod conflict. |
|
The primary side effect here is that new nodes might need additional 30s until routes are setup, if the project already had wonky routes. |
apricote
pushed a commit
that referenced
this pull request
Jul 8, 2024
🤖 I have created a release *beep* *boop* --- ## [1.20.0](v1.19.0...v1.20.0) (2024-07-08) ### Features * add support & tests for Kubernetes 1.29 ([#600](#600)) ([e8fabda](e8fabda)) * add support & tests for Kubernetes 1.30 ([#679](#679)) ([0748b6e](0748b6e)) * drop tests for kubernetes v1.25 ([#597](#597)) ([58261ec](58261ec)) * drop tests for kubernetes v1.26 ([#680](#680)) ([9c4be01](9c4be01)) * emit Kubernetes events for error conditions ([#598](#598)) ([e8f9199](e8f9199)) * **helm,manifests:** only specify container args instead of command ([#691](#691)) ([2ba4058](2ba4058)) * **helm:** allow setting affinity for deployment ([#686](#686)) ([1a8ea95](1a8ea95)) * read HCLOUD_TOKEN from file ([#652](#652)) ([a4343b8](a4343b8)) ### Bug Fixes * **routes:** many requests for outdated routes by rate limiting ([#675](#675)) ([e283b7d](e283b7d)) --- This PR was generated with [Release Please](https://github.com/googleapis/release-please). See [documentation](https://github.com/googleapis/release-please#release-please).
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #673
In
routes.ListRoutes()we have to find the matching server/node for every route in the network. We find the server by utilizing a cache that maps every private IP to the corresponding server.This cache has a feature that refreshes the list of servers if an entry can not be found. This is sensible, as the server might have been just created. This is also fatal, as this refresh happens for every single cache access. If there are a lot of routes in the network that do not belong to any server, we refresh the cache many times for each
ListRoutes(). This is even more serious, asListRoutes()is being called every 10-30 seconds (see #395).This commit introduces a
rate.Limiterin theAllServersCachewhich only allows the refresh to happen every 30 seconds.