You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
- The Blueprint `modelId` uses a raw foundation model ID (e.g. `anthropic.claude-opus-4-20250514-v1:0`)
116
116
- Fix: change to the inference profile ID (e.g. `us.anthropic.claude-opus-4-20250514-v1:0`), update DynamoDB via redeploy
117
117
118
+
**503 "Too many connections" / task completes with 0 tokens after long duration:**
119
+
- Bedrock is throttling model invocations. The agent retries for minutes then gives up.
120
+
- Symptoms: task runs for 10-15 minutes, completes with `COMPLETED` status but 0 tokens, 0 cost, no PR, `disk_delta: 0 B`
121
+
- Diagnosis:
122
+
1. Check application logs for `"text": "API Error: 503 Too many connections"`
123
+
2.**Check what model_id is actually being passed** — the DynamoDB record may have a stale model override:
124
+
```bash
125
+
aws dynamodb get-item \
126
+
--table-name <RepoTableName> \
127
+
--key '{"repo": {"S": "owner/repo"}}' \
128
+
--query 'Item.model_id' --output text
129
+
```
130
+
- Causes:
131
+
- **Stale model_id in DynamoDB** (most common) — the Blueprint `onUpdate` only sets fields present in props; removing a `modelId` prop does NOT remove the field from DynamoDB. The task keeps using the old model.
132
+
- Bedrock service-level throttling for the specific model (especially Opus 4 which has limited availability)
133
+
- Account quota limits reached
134
+
- Fix:
135
+
1. **Check and fix the DynamoDB record first** — remove stale `model_id`if present:
136
+
```bash
137
+
aws dynamodb update-item \
138
+
--table-name <RepoTableName> \
139
+
--key '{"repo": {"S": "owner/repo"}}' \
140
+
--update-expression "REMOVE model_id"
141
+
```
142
+
2. If model_id is correct, wait and retry — throttling is often transient
143
+
3. Switch to a model with higher availability (Sonnet 4.6 > Opus 4 > Haiku)
144
+
4. Request a Bedrock quota increase for`InvokeModel` RPM on your model
145
+
118
146
**`task_timed_out`:**
119
147
- 9-hour maximum exceeded
120
148
- Consider reducing scope or increasing `max_turns`for complex tasks
0 commit comments