Recommended Thresholds
- Max batch size: 10 profiles per wave for initial rollout.
- Failure stop trigger: > 20% in any 10-minute window.
- Retry cap: 1 immediate retry, then quarantine.
- Timeout classes split by startup, runtime, and cleanup.
Multilogin X Script Runner Playbook
This guide formalizes script-runner usage into a controlled batch workflow with payload contracts, safety checks, and incident-aware execution limits.
Payload Contract
Minimum Request Shape
{
"script_file": "profile.py",
"profile_ids": [
{
"profile_id": "2b91e901-4606-46fc-af20-f93a8865a7ff",
"is_headless": true
}
]
}Keep schema strict so failed payloads are rejected early instead of poisoning whole batches.
Execution Safety
Batch Strategy That Does Not Collapse
Step 1: Validate payload schema and profile ownership before queueing.
Step 2: Split workloads into small cohorts with max concurrency limits.
Step 3: Track each job by trace_id, profile_id, and script_version.
Step 4: Stop or quarantine a cohort when failure ratio crosses threshold.
Step 5: Save run evidence and summarize pass-fail matrix for decision pages.
Evidence Fields
- trace_id and batch_id
- profile_id and script_file
- headless flag
- duration_ms and end_state
- failure_class and rollback_action
Pseudo Control Loop
def run_batch(batch):
results = []
for job in batch:
if should_pause_batch(results):
quarantine_remaining(batch, results)
break
result = run_script_job(job)
results.append(result)
publish_batch_summary(results)
return resultsPause and quarantine logic is more valuable than blind retries under unstable conditions.
Failure Matrix
Common Script Runner Problems
| Problem | Likely cause | Fast action |
|---|---|---|
| Batch-wide crash wave | Single malformed payload pushed to full cohort | Introduce schema gate and canary cohort before full run. |
| Random timeout spikes | No per-stage timeout budget | Split timeout by startup, run, and stop stages. |
| Orphan sessions after failure | No cleanup enforcement | Force stop in finally and track cleanup result codes. |
| No audit trace for incidents | Weak logging schema | Use mandatory trace_id and batch summary artifacts. |
Affiliate Connection
Use Batch Reliability to Qualify Buyer Traffic
Readers trust recommendations more when they see controlled run evidence. Publish your pass thresholds, then route to commercial pages.
FAQ
Script Runner Questions
What is the biggest batch risk?
Failure propagation from one invalid payload or profile state to the whole cohort.
Should headless mode always be on?
No, use it only where your workload and repeated checks confirm stable behavior.
How does this improve affiliate conversion quality?
Evidence-backed reliability attracts better-fit buyers and reduces low-confidence traffic.