fix(server): resolve memory leaks causing OOM in k8s deployment

- usage/logger_plugin: cap modelStats.Details at 1000 entries per model
- cache/signature_cache: add background cleanup for expired sessions (10 min)
- management/handler: add background cleanup for stale IP rate-limit entries (1 hr)
- executor/cache_helpers: add mutex protection and TTL cleanup for codexCacheMap (15 min)
- executor/codex_executor: use thread-safe cache accessors

Add reproduction tests demonstrating leak behavior before/after fixes.

Amp-Thread-ID: https://ampcode.com/threads/T-019ba0fc-1d7b-7338-8e1d-ca0520412777
Co-authored-by: Amp <amp@ampcode.com>
This commit is contained in:
hemanta212
2026-01-09 13:33:46 +05:45
parent d47b7dc79a
commit 47dacce6ea
7 changed files with 526 additions and 6 deletions

View File

@@ -26,11 +26,17 @@ const (
// MinValidSignatureLen is the minimum length for a signature to be considered valid
MinValidSignatureLen = 50
// SessionCleanupInterval controls how often stale sessions are purged
SessionCleanupInterval = 10 * time.Minute
)
// signatureCache stores signatures by sessionId -> textHash -> SignatureEntry
var signatureCache sync.Map
// sessionCleanupOnce ensures the background cleanup goroutine starts only once
var sessionCleanupOnce sync.Once
// sessionCache is the inner map type
type sessionCache struct {
mu sync.RWMutex
@@ -45,6 +51,9 @@ func hashText(text string) string {
// getOrCreateSession gets or creates a session cache
func getOrCreateSession(sessionID string) *sessionCache {
// Start background cleanup on first access
sessionCleanupOnce.Do(startSessionCleanup)
if val, ok := signatureCache.Load(sessionID); ok {
return val.(*sessionCache)
}
@@ -53,6 +62,40 @@ func getOrCreateSession(sessionID string) *sessionCache {
return actual.(*sessionCache)
}
// startSessionCleanup launches a background goroutine that periodically
// removes sessions where all entries have expired.
func startSessionCleanup() {
go func() {
ticker := time.NewTicker(SessionCleanupInterval)
defer ticker.Stop()
for range ticker.C {
purgeExpiredSessions()
}
}()
}
// purgeExpiredSessions removes sessions with no valid (non-expired) entries.
func purgeExpiredSessions() {
now := time.Now()
signatureCache.Range(func(key, value any) bool {
sc := value.(*sessionCache)
sc.mu.Lock()
// Remove expired entries
for k, entry := range sc.entries {
if now.Sub(entry.Timestamp) > SignatureCacheTTL {
delete(sc.entries, k)
}
}
isEmpty := len(sc.entries) == 0
sc.mu.Unlock()
// Remove session if empty
if isEmpty {
signatureCache.Delete(key)
}
return true
})
}
// CacheSignature stores a thinking signature for a given session and text.
// Used for Claude models that require signed thinking blocks in multi-turn conversations.
func CacheSignature(sessionID, text, signature string) {