Open
Conversation
JaneLiuL
requested review from
CH3CHO,
erasernoob,
johnlanni and
rinfx
as code owners
|
undertaker86001 seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account. You have signed the CLA already but the status is still pending? Let us recheck it. |
kaori-seasons
force-pushed
the
feature/token
branch
from
ff1eeaa to
e719cad
Compare
JaneLiuL
changed the title
CH3CHO
reviewed
Jan 16, 2026
| normalizedType := strings.TrimSpace(strings.ToLower(model)) | ||
|
|
||
| // 处理别名（兼容不同命名习惯） | ||
| switch normalizedType { |
Author
There was a problem hiding this comment.
是需要处理如果model比如用户输入错误的model 要兼容对吗？是否需要加入模糊匹配？
CH3CHO
reviewed
Jan 16, 2026
| defaultConfigDisable: false | ||
| ``` | ||
|
|
||
| ### 4.6.3 功能验证 |
Author
There was a problem hiding this comment.
本地的一些笔记，写readme的时候不想重复写把本地笔记拷贝了，会更正的
| go run ./ | ||
| ``` | ||
|
|
||
| 跟着官网https://higress.cn/zh-cn/blog/e2e-debug/测试插件 |
|
|
||
| 注意：插件在真实运行时依赖 proxy-wasm hostcalls（例如定义/记录指标、获取 HTTP header 等）。直接在本地通过 `go run` 或 `go test` 运行会遇到 hostcall 不可用的问题（会触发 panic）。为方便本地开发，本仓库做了以下改进： | ||
|
|
||
| - 在 `main.go` 中，将指标抽象为 `metricCounter` 并提供 `noopCounter`，在非 Wasm 环境下跳过调用真实 hostcalls，避免 panic。 |
| 如有问题，请在仓库中创建 issue，或直接在 PR 中指明你遇到的问题与预期行为。 | ||
|
|
||
| 创建WasmPlugin资源配置： | ||
| 确保插件执行在 AI-Cache 插件之后（通过priority配置，数值越大执行越晚）； |
Collaborator
There was a problem hiding this comment.
这部分最好放在配置那边，也最好注明推荐的执行阶段和优先级。现在太不显眼了
| @@ -0,0 +1,180 @@ | |||
| # Token Statistics Plugin | |||
|
|
|||
| 该目录实现了 Higress 的 `token-statistics` wasm 插件，用于从各类 AI 服务响应中提取 token 使用量并导出到日志/指标，以便进行成本统计与监控。 | |||
Collaborator
There was a problem hiding this comment.
这个插件与 ai-statistics 插件的关系是什么呢，是否可以同时使用？
CH3CHO
reviewed
Jan 16, 2026
added 2 commits
January 19, 2026 09:54
…ken-statistics Key improvements: - Restore full Prometheus metrics verification logic - Increase retry count from 10 to 15 times (45s total timeout) - Increase retry interval from 2s to 3s for better stability - Add detailed error logging and troubleshooting hints - Implement graceful degradation when Prometheus unavailable - Restore complete exporter and dimensions configuration Metrics verified: - input_tokens_total - output_tokens_total - total_tokens_total - Aggregate queries across all models
Root cause: - geo-ip plugin initialization fails due to resource contention with token-statistics plugin in CI - Default FAIL_CLOSE strategy causes all requests to return 500 error - Memory pressure from loading 50k entries exceeds WASM VM limits Fixes: 1. Add fail_strategy: FAIL_OPEN to geo-ip WasmPlugin config - Enables graceful degradation for non-critical plugin - Prevents 500 errors when initialization fails 2. Reduce MaxEntriesForCI from 50000 to 10000 - Decreases memory footprint (10MB -> 2MB) - Reduces initialization time (~5s -> ~1s) - Test IPs are hardcoded, no impact on test coverage Impact: - Resolves CI timeout and 500 error issues - Improves plugin reliability in resource-constrained environments
…ailure Critical fix for CI 500 error: Root cause: - parseConfig and ReadGeoIpDataProgressively lack panic recovery - Panic in OnPluginStart phase causes OnPluginStartStatusFailed - Even with fail_strategy: FAIL_OPEN, panic bypasses the mechanism - Result: All requests return 500 error Fixes: 1. Add panic recovery to parseConfig with named return error - Captures panic and converts to error - Allows OnPluginStart to handle error gracefully 2. Add panic recovery to ReadGeoIpDataProgressively - Prevents initialization panic from failing plugin - Works with fail_strategy: FAIL_OPEN for graceful degradation 3. Add IPTree nil check for additional safety 4. Add fmt import for fmt.Errorf Technical details: - Changed function signatures to use named return (err error) - defer+recover converts panic to error before OnPluginStart sees it - fail_strategy: FAIL_OPEN now works correctly with error returns - MaxEntriesForCI=10000 reduces resource pressure Impact: - Prevents 500 errors in resource-constrained CI environments - Enables graceful plugin degradation when initialization fails - Compatible with existing fail_strategy: FAIL_OPEN configuration
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Ⅰ. Describe what this PR did
fix #3214
Ⅱ. Does this pull request fix one issue?
fix #3214
Ⅲ. Why don't you add test cases (unit test/integration test)?
Ⅳ. Describe how to verify it
run e2e for verify
Ⅴ. Special notes for reviews
Ⅵ. AI Coding Tool Usage Checklist (if applicable)
no ai coding tools
Please check all applicable items:
design/directory in the plugin folderdesign/directorywill add design directory later
AI Coding Prompts (for regular updates)
AI Coding Summary