jianf/ModelRouter
1
0
Fork 0
mirror of https://gitcode.com/JianFeeeee/ModelRouter.git synced 2026-09-20 00:48:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: restart wipes stats history — LoadAudit replayed only the last 3000 lines (92% of which are 3s access events) and fed access rows into aggregates; now replays every real request into totals/quota windows, skips event rows, tolerates >64KB lines; +TestLoadAuditFullReplay

Browse Source
This commit is contained in:
JianFeeeee
2026-08-11 08:42:13 +08:00
parent 6221502adc
commit e8ef73321c
3 changed files with 82 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
internal/gateway/stats.go
View File

@ -135,21 +135,30 @@ func incStatus(a *Stat, name string, r Req) {
func (s *Stats) LoadAudit(path string) {
f, err := os.Open(path)
if err == nil {
defer f.Close()
s.mu.Lock()
defer s.mu.Unlock()
var recs []Req
// tolerate long error summaries / oversized junk lines
sc := bufio.NewScanner(f)
sc.Buffer(make([]byte, 64*1024), 16*1024*1024)
for sc.Scan() {
var r Req
if json.Unmarshal(sc.Bytes(), &r) == nil {
recs = append(recs, r)
if json.Unmarshal(sc.Bytes(), &r) != nil || r.Type == "" {
continue // access/event rows (obj) and malformed lines are no requests
}
// replay EVERY request row into the aggregates so totals and
// quota windows survive restarts; only the ring buffer view stays
// capped at maxRecs.
s.aggregateLocked(r)
recs = append(recs, r)
}
_ = f.Close()
if len(recs) > s.maxRecs {
recs = recs[len(recs)-s.maxRecs:]
}
for _, r := range recs {
s.Record(r)
if n := len(recs); n > s.maxRecs {
recs = append([]Req(nil), recs[n-s.maxRecs:]...)
}
s.recs = recs
s.auditPath = path
return
}
s.mu.Lock()
s.auditPath = path
@ -160,6 +169,20 @@ func (s *Stats) LoadAudit(path string) {
func (s *Stats) Record(r Req) {
s.mu.Lock()
defer s.mu.Unlock()
s.aggregateLocked(r)
s.recs = append(s.recs, r)
if len(s.recs) > s.maxRecs {
s.recs = s.recs[len(s.recs)-s.maxRecs:]
}
if s.auditPath != "" {
s.rotateAuditLocked()
appendAuditLine(s.auditPath, r)
}
}
// aggregateLocked folds r into every aggregate row and the quota window
// bucket. Caller must hold s.mu.
func (s *Stats) aggregateLocked(r Req) {
inc(s.byKey, r.Key, r)
if r.Model != "" {
inc(s.byModel, r.Model, r)
@ -180,13 +203,12 @@ func (s *Stats) Record(r Req) {
}
inc(ks, r.Source, r)
if r.Status > 0 {
name := strconv.Itoa(r.Status)
a := s.byStatus[r.Status]
if a == nil {
a = &Stat{}
s.byStatus[r.Status] = a
}
incStatus(a, name, r)
incStatus(a, strconv.Itoa(r.Status), r)
}
// window bucket for quota enforcement (per source-model pair, per unix hour)
tok := r.Prompt + r.Compl
@ -210,14 +232,6 @@ func (s *Stats) Record(r Req) {
}
}
}
s.recs = append(s.recs, r)
if len(s.recs) > s.maxRecs {
s.recs = s.recs[len(s.recs)-s.maxRecs:]
}
if s.auditPath != "" {
s.rotateAuditLocked()
appendAuditLine(s.auditPath, r)
}
}
// rotateAuditLocked renames the audit file to <path>.<unix>.old once it