fix(api): Hide internal attributes in events and cross-trace queries by nsdeschenes · Pull Request #116094 · getsentry/sentry

nsdeschenes · 2026-05-22T13:48:42Z

Summary

Apply visibility checks to organization_events and organization_events_timeseries endpoints — queries on hidden internal attributes return empty results for non-staff users
Refactor get_cross_trace_queries to accept the full SearchResolver instead of config + params, propagating hidden attribute markers from cross-trace sub-resolvers back to the primary resolver
Add has_hidden_api_attributes() short-circuit in both run_table_query and run_bulk_table_query to avoid dispatching RPCs
Derive the visibility item type from the resolver's column definitions rather than the primary request type, so cross-trace sub-resolvers use the correct item type
Return properly structured empty timeseries responses (with field metadata) when attributes are hidden

Depends on #116091

Test plan

New tests in test_organization_events_span_indexed.py, test_organization_events_timeseries_spans.py, test_organization_events_cross_trace.py, and test_rpc_dataset_common.py
Coverage for log resolvers receiving a spans visibility config in test_ourlogs.py

Closes TODO

Add a shared helper for hiding internal Sentry convention attributes from API surfaces and let SearchResolver track attributes hidden by API visibility configuration. This keeps default resolver behavior unchanged unless an API caller opts into visibility enforcement.

Apply API attribute visibility checks when the events endpoint queries RPC datasets. This prevents non-staff users from selecting internal attributes with field params and receiving their values. Co-Authored-By: OpenAI Codex <noreply@openai.com>

Apply the same API attribute visibility checks to the events-timeseries RPC path. Return an empty timeseries response instead of querying hidden internal attributes for non-staff users. Co-Authored-By: OpenAI Codex <noreply@openai.com>

Return empty timeseries metadata when API visibility hides requested attributes, and type resolver visibility options explicitly so backend typing accepts the config calls. Co-Authored-By: Codex <noreply@openai.com>

Rename the hidden-attribute metadata variable so mypy does not see two typed definitions for final_meta in the same function scope. Co-Authored-By: Codex <codex@openai.com>

Propagate hidden API attribute markers from cross-trace resolvers back to the primary resolver so table and bulk table queries return empty results instead of dispatching RPCs. Resolve cross-trace query visibility using the target trace item type for span, log, metric, and occurrence filters. Add regression coverage for non-staff cross-trace filters on internal span and log attributes. Co-Authored-By: Codex <noreply@openai.com>

Add regression coverage that cross-trace occurrence filters resolve API attribute visibility with the occurrence item type instead of the primary spans item type. Co-Authored-By: Codex <noreply@openai.com>

Use the trace item type from the resolver's column definitions for API attribute visibility checks. This keeps dataset-specific resolvers from applying a mismatched primary request item type. Add coverage for log resolvers receiving a spans visibility config. Co-Authored-By: Codex <noreply@openai.com>

cursor

Cursor Bugbot has reviewed your changes and found 1 potential issue.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit e480b18. Configure here.}

cursor · 2026-05-22T13:54:34Z

            request_context_pairs.append(
                (query.name, table_request, cls.build_rpc_table_row_context(query))
            )
-        responses = snuba_rpc.table_rpc(


Table RPC skips hidden short-circuit

High Severity

This change adds has_hidden_api_attributes() guards before Snuba in bulk table and plain timeseries paths, but _run_table_query still always calls snuba_rpc.table_rpc after resolution. Organization events and top-events timeseries table steps use that path, so non-staff queries touching hidden internal fields or cross-trace filters can still return real data instead of the empty results the new tests expect.

^{Reviewed by Cursor Bugbot for commit e480b18. Configure here.}

linear-code · 2026-05-22T13:54:39Z

EXP-966

github-actions · 2026-05-22T13:56:49Z

Backend Test Failures

Failures on f7e99dd in this run:

tests/snuba/api/endpoints/test_organization_events_cross_trace.py::OrganizationEventsCrossTraceEndpointTest::test_cross_trace_query_with_internal_log_attribute_is_staff_only

— log

[gw1] linux -- Python 3.13.1 /home/runner/work/sentry/sentry/.venv/bin/python3
tests/snuba/api/endpoints/test_organization_events_cross_trace.py:195: in test_cross_trace_query_with_internal_log_attribute_is_staff_only
    assert response.data["data"] == []
E   AssertionError: assert [{'count()': ...oo]': 'five'}] == []
E     
E     Left contains one more item: �[0m{�[33m'�[39;49;00m�[33mcount()�[39;49;00m�[33m'�[39;49;00m: �[94m1.0�[39;49;00m, �[33m'�[39;49;00m�[33mtags[foo]�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33mfive�[39;49;00m�[33m'�[39;49;00m}�[90m�[39;49;00m
E     
E     Full diff:
E     �[0m�[91m- []�[39;49;00m�[90m�[39;49;00m
E     �[92m+ [�[39;49;00m�[90m�[39;49;00m
E     �[92m+     {�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'count()': 1.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'tags[foo]': 'five',�[39;49;00m�[90m�[39;49;00m
E     �[92m+     },�[39;49;00m�[90m�[39;49;00m
E     �[92m+ ]�[39;49;00m�[90m�[39;49;00m

tests/snuba/api/endpoints/test_organization_events_span_indexed.py::OrganizationEventsSpansEndpointTest::test_sentry_internal_field_values_are_staff_only

— log

[gw0] linux -- Python 3.13.1 /home/runner/work/sentry/sentry/.venv/bin/python3
tests/snuba/api/endpoints/test_organization_events_span_indexed.py:4579: in test_sentry_internal_field_values_are_staff_only
    assert response.data["data"] == []
E   AssertionError: assert [{'__sentry_i...ount()': 1.0}] == []
E     
E     Left contains one more item: �[0m{�[33m'�[39;49;00m�[33m__sentry_internal_test�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33minternal_value�[39;49;00m�[33m'�[39;49;00m, �[33m'�[39;49;00m�[33mcount()�[39;49;00m�[33m'�[39;49;00m: �[94m1.0�[39;49;00m}�[90m�[39;49;00m
E     
E     Full diff:
E     �[0m�[91m- []�[39;49;00m�[90m�[39;49;00m
E     �[92m+ [�[39;49;00m�[90m�[39;49;00m
E     �[92m+     {�[39;49;00m�[90m�[39;49;00m
E     �[92m+         '__sentry_internal_test': 'internal_value',�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'count()': 1.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+     },�[39;49;00m�[90m�[39;49;00m
E     �[92m+ ]�[39;49;00m�[90m�[39;49;00m

tests/snuba/api/endpoints/test_organization_events_cross_trace.py::OrganizationEventsCrossTraceEndpointTest::test_cross_trace_query_with_internal_span_attribute_is_staff_only

— log

[gw1] linux -- Python 3.13.1 /home/runner/work/sentry/sentry/.venv/bin/python3
tests/snuba/api/endpoints/test_organization_events_cross_trace.py:145: in test_cross_trace_query_with_internal_span_attribute_is_staff_only
    assert response.data["data"] == []
E   AssertionError: assert [{'count()': ...oo]': 'five'}] == []
E     
E     Left contains one more item: �[0m{�[33m'�[39;49;00m�[33mcount()�[39;49;00m�[33m'�[39;49;00m: �[94m1.0�[39;49;00m, �[33m'�[39;49;00m�[33mtags[foo]�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33mfive�[39;49;00m�[33m'�[39;49;00m}�[90m�[39;49;00m
E     
E     Full diff:
E     �[0m�[91m- []�[39;49;00m�[90m�[39;49;00m
E     �[92m+ [�[39;49;00m�[90m�[39;49;00m
E     �[92m+     {�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'count()': 1.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'tags[foo]': 'five',�[39;49;00m�[90m�[39;49;00m
E     �[92m+     },�[39;49;00m�[90m�[39;49;00m
E     �[92m+ ]�[39;49;00m�[90m�[39;49;00m

tests/snuba/api/endpoints/test_organization_events_timeseries_spans.py::OrganizationEventsStatsSpansMetricsEndpointTest::test_sentry_internal_groupby_values_are_staff_only

— log

[gw1] linux -- Python 3.13.1 /home/runner/work/sentry/sentry/.venv/bin/python3
tests/snuba/api/endpoints/test_organization_events_timeseries_spans.py:168: in test_sentry_internal_groupby_values_are_staff_only
    assert response.data["timeSeries"] == []
E   AssertionError: assert [{'groupBy': ...': 'count()'}] == []
E     
E     Left contains one more item: �[0m{�[33m'�[39;49;00m�[33mgroupBy�[39;49;00m�[33m'�[39;49;00m: [{�[33m'�[39;49;00m�[33mkey�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33m__sentry_internal_test�[39;49;00m�[33m'�[39;49;00m, �[33m'�[39;49;00m�[33mvalue�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33minternal_value�[39;49;00m�[33m'�[39;49;00m}], �[33m'�[39;49;00m�[33mmeta�[39;49;00m�[33m'�[39;49;00m: {�[33m'�[39;49;00m�[33mdataScanned�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33mfull�[39;49;00m�[33m'�[39;49;00m, �[33m'�[39;49;00m�[33minterval�[39;49;00m�[33m'�[39;49;00m... �[94mNone�[39;49;00m, ...}, {�[33m'�[39;49;00m�[33mconfidence�[39;49;00m�[33m'�[39;49;00m: �[94mNone�[39;49;00m, �[33m'�[39;49;00m�[33mincomplete�[39;49;00m�[33m'�[39;49;00m: �[94mFalse�[39;49;00m, �[33m'�[39;49;00m�[33msampleCount�[39;49;00m�[33m'�[39;49;00m: �[94m0�[39;49;00m, �[33m'�[39;49;00m�[33msampleRate�[39;49;00m�[33m'�[39;49;00m: �[94mNone�[39;49;00m, ...}], �[33m'�[39;49;00m�[33myAxis�[39;49;00m�[33m'�[39;49;00m: �[33m'�[39;49;00m�[33mcount()�[39;49;00m�[33m'�[39;49;00m}�[90m�[39;49;00m
E     
E     Full diff:
E     �[0m�[91m- []�[39;49;00m�[90m�[39;49;00m
E     �[92m+ [�[39;49;00m�[90m�[39;49;00m
E     �[92m+     {�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'groupBy': [�[39;49;00m�[90m�[39;49;00m
E     �[92m+             {�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'key': '__sentry_internal_test',�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'value': 'internal_value',�[39;49;00m�[90m�[39;49;00m
E     �[92m+             },�[39;49;00m�[90m�[39;49;00m
E     �[92m+         ],�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'meta': {�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'dataScanned': 'full',�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'interval': 3600000,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'isOther': False,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'order': 0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'valueType': 'integer',�[39;49;00m�[90m�[39;49;00m
E     �[92m+             'valueUnit': None,�[39;49;00m�[90m�[39;49;00m
E     �[92m+         },�[39;49;00m�[90m�[39;49;00m
E     �[92m+         'values': [�[39;49;00m�[90m�[39;49;00m
E     �[92m+             {�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'confidence': 'high',�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'incomplete': False,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleCount': 1,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleRate': 1.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'timestamp': 1779357600000,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'value': 1.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             },�[39;49;00m�[90m�[39;49;00m
E     �[92m+             {�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'confidence': None,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'incomplete': False,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleCount': 0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleRate': None,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'timestamp': 1779361200000,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'value': 0.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             },�[39;49;00m�[90m�[39;49;00m
E     �[92m+             {�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'confidence': None,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'incomplete': False,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleCount': 0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'sampleRate': None,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'timestamp': 1779364800000,�[39;49;00m�[90m�[39;49;00m
E     �[92m+                 'value': 0.0,�[39;49;00m�[90m�[39;49;00m
E     �[92m+             },�[39;49;00m�[90m�[39;49;00m
... (28 more lines)

nsdeschenes · 2026-05-27T17:33:14Z

Closing out in favour of a clean slate.

nsdeschenes and others added 8 commits May 22, 2026 10:45

fix(api): Handle hidden EAP attributes in RPC queries

1d3d45a

Return empty timeseries metadata when API visibility hides requested attributes, and type resolver visibility options explicitly so backend typing accepts the config calls. Co-Authored-By: Codex <noreply@openai.com>

fix(snuba): Avoid duplicate metadata annotation

459b3b0

Rename the hidden-attribute metadata variable so mypy does not see two typed definitions for final_meta in the same function scope. Co-Authored-By: Codex <codex@openai.com>

test(snuba): Cover cross-trace visibility item types

b4dcf7b

Add regression coverage that cross-trace occurrence filters resolve API attribute visibility with the occurrence item type instead of the primary spans item type. Co-Authored-By: Codex <noreply@openai.com>

github-actions Bot added the Scope: Backend Automatically applied to PRs that change backend components label May 22, 2026

nsdeschenes mentioned this pull request May 22, 2026

feat(attributes): Hide internal attributes from responses #116084

Closed

cursor Bot reviewed May 22, 2026

View reviewed changes

Base automatically changed from nd/feat-attributes-visibility-core to master May 27, 2026 13:24

nsdeschenes closed this May 27, 2026

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

fix(api): Hide internal attributes in events and cross-trace queries#116094

fix(api): Hide internal attributes in events and cross-trace queries#116094
nsdeschenes wants to merge 8 commits into
masterfrom
nd/feat-attributes-visibility-events

nsdeschenes commented May 22, 2026

Uh oh!

cursor Bot left a comment

Uh oh!

cursor Bot May 22, 2026

Uh oh!

linear-code Bot commented May 22, 2026

Uh oh!

github-actions Bot commented May 22, 2026 •

edited

Loading

Uh oh!

nsdeschenes commented May 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Uh oh!

Conversation

nsdeschenes commented May 22, 2026

Summary

Test plan

Uh oh!

cursor Bot left a comment

Choose a reason for hiding this comment

Uh oh!

cursor Bot May 22, 2026

Choose a reason for hiding this comment

Table RPC skips hidden short-circuit

Uh oh!

linear-code Bot commented May 22, 2026

Uh oh!

github-actions Bot commented May 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Backend Test Failures

Uh oh!

nsdeschenes commented May 27, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

github-actions Bot commented May 22, 2026 •

edited

Loading