github/awesome-kubernetes
1
0
Fork 0
mirror of https://github.com/nubenetes/awesome-kubernetes.git synced 2026-07-12 18:00:37 +00:00
Code Issues Packages Projects Releases Wiki Activity

build: inject detailed telemetry logging for precise quota diagnostics

Browse Source 
- Added payload size estimation (tokens) before requests.
- Added exact Token consumption (Prompt/Candidates) on success.
- Added exact error messages (response.text) on 429 and 50x errors to diagnose quota exhaustion limits (RPM vs TPM).
This commit is contained in:
Nubenetes Bot
2026-05-21 13:44:13 +02:00
parent 6e191c8ab2
commit 9bb60eef29
1 changed files with 12 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
src/gemini_utils.py
View File

@@ -334,6 +334,11 @@ async def call_gemini_with_retry(prompt: str, response_format: str = "json", max
"contents": [{"parts": [{"text": prompt}]}],
"tools": [{"google_search_retrieval": {}}] if use_grounding else []
}
# TELEMETRY: Log Payload Size
payload_size = len(json.dumps(payload))
log_event(f" [TELEMETRY] Attempting Key {current_idx+1} | Model: {model} | Payload: ~{payload_size//4} tokens")
response = await client.post(api_url, json=payload, timeout=60)
resp_json = {}
@@ -344,7 +349,7 @@ async def call_gemini_with_retry(prompt: str, response_format: str = "json", max
SESSION_TRACKER.track_call(current_idx, model, response.status_code, usage, role=role)
if response.status_code == 200:
log_event(f" [AI] Success: Key {current_idx+1} | Model: {model} | Role: {role}")
log_event(f" [AI] Success: Key {current_idx+1} | Model: {model} | Role: {role} | Tokens: P={usage.get('promptTokenCount',0)} C={usage.get('candidatesTokenCount',0)}")
CURRENT_KEY_INDEX = current_idx
if 'candidates' in resp_json and resp_json['candidates']:
text_resp = resp_json['candidates'][0]['content']['parts'][0]['text']
@@ -370,6 +375,9 @@ async def call_gemini_with_retry(prompt: str, response_format: str = "json", max
elif response.status_code == 429:
consecutive_429s += 1
# TELEMETRY: Log Exact Rejection Reason
log_event(f" [TELEMETRY] 429 Details: {response.text[:200]}")
# 2. ADAPTIVE TIERING: Mark this specific model as throttled
throttle_duration = 30 if "pro" in model else 15
THROTTLED_MODELS[f"{current_idx}_{model}"] = time.time() + throttle_duration
@@ -388,13 +396,16 @@ async def call_gemini_with_retry(prompt: str, response_format: str = "json", max
diagnostics.add_attempt(model, 404, "Not Found")
break
elif response.status_code in [500, 503, 504]:
log_event(f" [TELEMETRY] Server Error {response.status_code}: {response.text[:200]}")
diagnostics.add_attempt(model, response.status_code, "Server Error")
continue
else:
log_event(f" [TELEMETRY] Unknown Error {response.status_code}: {response.text[:200]}")
diagnostics.add_attempt(model, response.status_code, "API Error", response.text)
break
except Exception as e:
log_event(f" [TELEMETRY] Exception during request: {e}")
SESSION_TRACKER.track_call(current_idx, model, 0, {}, role=role)
diagnostics.add_attempt(model, 0, str(e))
break