feat: add automatic key failover for AI Bridge Anthropic (#24836)

## Description

Adds automatic key failover for centralized Anthropic provider. When a key pool is configured, each upstream call walks the pool and tries keys in order until one succeeds or the pool is exhausted. Keys are marked **temporary** on 429 (with cooldown from `Retry-After`) and **permanent** on 401/403. Errors that aren't key-specific don't trigger failover. Each agentic-loop iteration gets its own fresh walker, so a tool-call continuation can fail over independently of the initial request.

BYOK is unchanged: BYOK requests run as a single attempt with no failover.

## Changes

- `config.Anthropic` carries a `KeyPool`. `Key` remains for BYOK X-Api-Key set per interception.
- Blocking interceptor: walks the pool, marks keys on key-specific failures, returns on first success or non-failover error.
- Streaming interceptor: per-iteration walker. Pre-stream failures fail over to the next key; mid-stream errors are relayed as SSE events.
- New `keypool` error types: `TransientExhaustionError` (carries soonest cooldown) and `ErrPermanentExhaustion`. Replace the prior `ErrAllKeysExhausted`.
- Error responses now consistently include the outer `"type": "error"` field.

## Related Issues

Related to: https://github.com/coder/internal/issues/1446
Related to: https://linear.app/codercom/issue/AIGOV-197/aibridge-automatic-key-failover-for-bridged-and-passthrough-routes

## Follow-up PRs

- Bedrock multi-key support.
- Refactor provider vs interceptor config separation.
- Record the actually-used key in the interception credential hint after failover.

> [!NOTE]
> Initially generated by Claude Opus 4.7, modified and reviewed by @ssncferreira
This commit is contained in:
Susana Ferreira
2026-05-07 14:57:44 +01:00
committed by GitHub
parent 273e828442
commit f1155ac4d7
17 changed files with 2313 additions and 121 deletions
+26 -9
View File
@@ -10,10 +10,12 @@ import (
"github.com/coder/coder/v2/aibridge"
"github.com/coder/coder/v2/aibridge/config"
"github.com/coder/coder/v2/aibridge/keypool"
"github.com/coder/coder/v2/coderd/tracing"
"github.com/coder/coder/v2/codersdk"
"github.com/coder/coder/v2/enterprise/aibridged"
"github.com/coder/coder/v2/enterprise/coderd"
"github.com/coder/quartz"
)
func newAIBridgeDaemon(coderAPI *coderd.API, providers []aibridge.Provider) (*aibridged.Server, error) {
@@ -88,16 +90,24 @@ func buildProviders(cfg codersdk.AIBridgeConfig) ([]aibridge.Provider, error) {
}
// Add legacy Anthropic provider if configured. Bedrock credentials
// alone are sufficient — an Anthropic API key is not required when
// alone are sufficient, an Anthropic API key is not required when
// using AWS Bedrock.
if cfg.LegacyAnthropic.Key.String() != "" || getBedrockConfig(cfg.LegacyBedrock) != nil {
if _, conflict := usedNames[aibridge.ProviderAnthropic]; conflict {
return nil, xerrors.Errorf("legacy CODER_AIBRIDGE_ANTHROPIC_KEY conflicts with indexed provider named %q; remove one or the other", aibridge.ProviderAnthropic)
}
var pool *keypool.Pool
if key := cfg.LegacyAnthropic.Key.String(); key != "" {
var err error
pool, err = keypool.New([]string{key}, quartz.NewReal())
if err != nil {
return nil, xerrors.Errorf("create legacy anthropic key pool: %w", err)
}
}
providers = append(providers, aibridge.NewAnthropicProvider(aibridge.AnthropicConfig{
Name: aibridge.ProviderAnthropic,
BaseURL: cfg.LegacyAnthropic.BaseURL.String(),
Key: cfg.LegacyAnthropic.Key.String(),
KeyPool: pool,
CircuitBreaker: cbConfig,
SendActorHeaders: cfg.SendActorHeaders.Value(),
}, getBedrockConfig(cfg.LegacyBedrock)))
@@ -110,14 +120,13 @@ func buildProviders(cfg codersdk.AIBridgeConfig) ([]aibridge.Provider, error) {
if name == "" {
name = p.Type
}
// Currently, only the first key is used, if any.
// TODO(ssncferreira): pass a keypool.Pool instead.
var key string
if len(p.Keys) > 0 {
key = p.Keys[0]
}
switch p.Type {
case aibridge.ProviderOpenAI:
// TODO(ssncferreira): pass a keypool.Pool instead.
var key string
if len(p.Keys) > 0 {
key = p.Keys[0]
}
providers = append(providers, aibridge.NewOpenAIProvider(aibridge.OpenAIConfig{
Name: name,
BaseURL: p.BaseURL,
@@ -127,10 +136,18 @@ func buildProviders(cfg codersdk.AIBridgeConfig) ([]aibridge.Provider, error) {
SendActorHeaders: cfg.SendActorHeaders.Value(),
}))
case aibridge.ProviderAnthropic:
var pool *keypool.Pool
if len(p.Keys) > 0 {
var err error
pool, err = keypool.New(p.Keys, quartz.NewReal())
if err != nil {
return nil, xerrors.Errorf("create anthropic key pool for provider %q: %w", name, err)
}
}
providers = append(providers, aibridge.NewAnthropicProvider(aibridge.AnthropicConfig{
Name: name,
BaseURL: p.BaseURL,
Key: key,
KeyPool: pool,
APIDumpDir: p.DumpDir,
CircuitBreaker: cbConfig,
SendActorHeaders: cfg.SendActorHeaders.Value(),