Intermittent Boomi failures are the hardest kind — the process works most of the time, then fails without an obvious pattern. That intermittency is itself a clue. Here's the diagnostic order that isolates the cause fastest, from a supply chain integration standpoint.
1. Separate connection failures from data failures
First establish which kind of failure you have. A connection-level failure (timeout, refused, auth) points at the endpoint or network; a data-level failure (mapping error, validation, missing field) points at the document. The process logs distinguish these clearly — start there before theorising.
2. For connection failures, check the endpoint under load
Intermittent connection errors often mean the target system (WMS, ERP, partner gateway) is occasionally slow or rejecting under load. Look at whether failures cluster around peak processing windows, batch jobs, or maintenance times, and whether timeouts are set tightly enough to trip on a slow-but-valid response.
3. For data failures, find the input that breaks the map
A process that handles 95% of documents but fails on the rest usually meets a data shape it wasn't built for:
- Unexpected or missing fields: an optional element that's usually present but occasionally absent, or an extra segment from one partner.
- Format variations: date formats, decimal separators, character encoding, or field lengths that vary by source.
- Volume edge cases: an unusually large document, an empty batch, or a single-line order when multi-line is assumed.
Compare a failing document against a succeeding one field by field — the difference is the trigger.
4. Review retry and error-handling behaviour
Check whether the process retries on failure and what happens to a document that exhausts retries. A silent retry that eventually succeeds can mask an underlying instability; a failure that drops a document without alerting means data loss you may not have noticed. Confirm the error path is doing what you expect.
5. Check for concurrency and locking
If multiple executions run at once, intermittent failures can come from two processes contending for the same record, file, or connection pool. Look at whether failures correlate with overlapping runs.
6. Use the process logs and document tracking
Boomi's execution logs and document tracking will show the exact step and error for each failed run. Grouping failures by error message almost always reveals the common cause behind what looked like random behaviour.
The faster way
Intermittent integration failures are exactly what SupplyChain Assist is built to reason through — describe the symptom, the flow, and what varies when it fails, and it walks the likely causes with the specific checks for each, instead of you chasing an intermittent fault alone.