junit xml testsuite parsing errors in CI: how to fix

TL;DR

The XML is malformed: bad encoding, unescaped characters, or truncated writes. Validate the XML locally, fix the producer (encoding, buffering), and re-run.

Error

Failed to parse JUnit XML: not well-formed (invalid token) at line 412

Steps

1. Validate: xmllint --noout TEST-*.xml. Expected: the exact line of the break.
2. Look at that line: usually a control character in a test name or output. Expected: the offending bytes found.
3. Fix the producer: set UTF-8 encoding, strip control chars from test names. Expected: clean output.
4. If truncated, the test process died mid-write; fix the crash first. Expected: complete files.
5. Re-run and validate again. Expected: CI parses it.

When to use

• CI report step fails on XML parsing.
• Dashboards show missing results.

When not to use

• Tests fail (the XML is fine; the tests are not).
• You are choosing a report format (separate decision).

Tool compatibility

• Any JUnit XML producer; xmllint for validation.

Variant phrasings

JUnit XML not well-formed

The general error; validate to find the line.

Invalid token in TEST xml

Encoding or control characters.

Why it happens

Test output contains arbitrary bytes (logs, names). XML is strict; one bad byte breaks the whole file.

Edge cases

• Parallel writers to one file corrupt it; write per-worker files and merge.
• --junitxml paths must exist; pytest does not create directories.
• CDATA sections help but do not fix truly binary output.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst__9CFFVDKZnbnejYwa4CPRQ