apache/beam · error · AssertionError

Failed to generate a unique UUID for schema after 100…

Error message

Failed to generate a unique UUID for schema after 100 tries! Registry contains {len(self.by_id)} schemas.

What it means

The schema registry assigns each schema a UUID by retrying uuid4() up to 100 times until it finds one not already in by_id. If after 100 tries none were unique, it raises this AssertionError. In practice this means either a corrupted/poisoned registry (non-random uuid4, monkeypatched uuid) or an astronomically unlikely collision — with real uuid4 this is effectively an internal invariant violation.

Solutions

1. Stop stubbing/mocking uuid.uuid4 to return a fixed value in this process
2. Inspect the registry (len(by_id)) for duplicate/poisoned ids and clear it: SCHEMA_REGISTRY entries shouldn't collide for distinct schemas
3. Upgrade Beam; retrying 100 uuid4 draws failing is virtually impossible with a healthy RNG

Example fix

// before
mock.patch('uuid.uuid4', return_value='fixed-id')  # 100 iterations all collide
// after
mock.patch('uuid.uuid4', side_effect=[str(uuid4()) for _ in range(100)])

if any(sid == str(uuid4()) for sid in registry.by_id):  # sanity check only
  ...

try:
  schema_id = registry.generate_new_id()
except AssertionError as e:
  registry.clear(); schema_id = registry.generate_new_id()

When it happens

Trigger: Calling `generate_new_id()` (indirectly via `SchemaRegistry.add` / creating a schema-aware type) while the registry's by_id dict is corrupted or uuid4 is patched to return a constant value present in the registry.

Common situations: Test suites monkeypatching uuid.uuid4 with deterministic stubs; custom UUID generators reused for many schemas; extremely large registries with a broken RNG.

Understand the failure class

Background: "This is a bug, please report it": internal invariant violations, unreachable panics, and SNH errors explained — this error's family across 47 libraries.

Related errors

• Failed assert: element
• Failed assert: [] == %r
• Failed assert: %r == %r
• Failed assert: Received element
• Failed assert: unmatched elements

AI-assisted analysis of apache/beam@12126d8942 (2026-09-13). Data as JSON: /api/errors/f2cc5131afbc4e6c. Report an issue: GitHub.

Appendix: source

Thrown at sdks/python/apache_beam/typehints/schema_registry.py:37

any backwards-compatibility guarantee.
"""

from uuid import uuid4


# Registry of typings for a schema by UUID
class SchemaTypeRegistry(object):
  def __init__(self):
    self.by_id = {}
    self.by_typing = {}  # currently not used

  def generate_new_id(self):
    for _ in range(100):
      schema_id = str(uuid4())
      if schema_id not in self.by_id:
        return schema_id

    raise AssertionError(
        "Failed to generate a unique UUID for schema after "
        f"100 tries! Registry contains {len(self.by_id)} "
        "schemas.")

  def add(self, typing, schema):
    if schema.id:
      self.by_id[schema.id] = (typing, schema)

  def load_registered_typings(self, by_id):
    for id, typing in by_id.items():
      if id not in self.by_id:
        self.by_id[id] = (typing, None)

  def get_registered_typings(self):
    # Used by save_main_session, as pb2.schema isn't picklable
    return {k: v[0] for k, v in self.by_id.items()}

  def get_typing_by_id(self, unique_id):

View on GitHub (pinned to 12126d8942)