feat(configuration-issues): Making processing_errors queryable

[Abdkhan14]

Summary

Registers processing_errors as a queryable EAP dataset so the events and events-stats APIs can query processing error data (e.g. for issue details charts). Used established patterns for other queryable datasets as a guide.

• search/eap/processing_errors/ module with column, attribute and aggregate function definitions
• New ProcessingErrors RPC dataset class in snuba/processing_errors_rpc.py supporting table, timeseries, and top-events queries
• Registers the dataset in types.py, constants.py, utils.py and enables it in the organization_events and organization_events_stats endpoints (including top-events support)
• Exposes attribute mappings via the attribute mappings API
• Adds ProcessingErrorTestCase mixin with helpers to create and store processing error items

Test Plan

• tests/sentry/search/eap/test_processing_errors.py — validates query parsing (filters, negation, in-filter, AND/OR, empty)
• tests/snuba/api/endpoints/test_organization_events_stats_processing_errors.py — validates count, zerofill, top events grouped by error_type, filtering, and empty top events

[github-actions]

Backend Test Failures

Failures on d51fdb9 in this run:

tests/sentry/api/endpoints/test_organization_attribute_mappings.py::OrganizationAttributeMappingsEndpointTest::test_get_all_mappings — log

tests/sentry/api/endpoints/test_organization_attribute_mappings.py:34: in test_get_all_mappings
    assert types_present == expected_types
E   AssertionError: assert {'logs', 'occ...metrics', ...} == {'logs', 'occ...time_results'}
E     
E     Extra items in the left set:
E     'processing_errors'
E     
E     Full diff:
E       {
E           'logs',
E           'occurrences',
E     +     'processing_errors',
E           'profiles',
E           'spans',
E           'tracemetrics',
E           'uptime_results',
E       }

[Abdkhan14]

@wedamija I've added a todo for the received attr from #109851. Will follow up when your pr is merged.

[cursor]

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

Autofix Details

Bugbot Autofix prepared a fix for the issue found in the latest run.

• ✅ Fixed: Parameter additional_queries silently dropped in table query
  • ProcessingErrors.run_table_query now forwards additional_queries into rpc_dataset_common.TableQuery, so caller-provided additional filters are no longer dropped.

Or push these changes by commenting:

@cursor push b11c4835be

Preview (b11c4835be)

diff --git a/src/sentry/snuba/processing_errors_rpc.py b/src/sentry/snuba/processing_errors_rpc.py
--- a/src/sentry/snuba/processing_errors_rpc.py
+++ b/src/sentry/snuba/processing_errors_rpc.py
@@ -46,6 +46,7 @@
                 sampling_mode=sampling_mode,
                 resolver=search_resolver or cls.get_resolver(params, config),
                 page_token=page_token,
+                additional_queries=additional_queries,
             ),
             params.debug,
         )

_{This Bugbot Autofix run was free. To enable autofix for future PRs, go to the Cursor dashboard.}

[wedamija]

The received field pr has merged, although idk if it even needs to be queryable. Doesn't hurt to add it though.

Lgtm, just a couple of nits

[wedamija]

Let's import up the top

[Abdkhan14]

This is done

[wedamija]

I added a method for storing these in https://github.com/getsentry/sentry/pull/109851/changes#diff-044678ee783c164a8acca43b4fa4a97e7402d231129865acf6a253e1670e922f, would be good to see if it works for your usecase so that we don't duplicate work

[Abdkhan14]

@wedamija store_processing_errors is replaceable by produce_and_store_eap_items, however the latter tests producer correctness, requires passing producer specific params that seem redundant for API tests that only need the data to exist for testing querying behaviour. Also requires more boilterplate for forming the event_data and errors schema like here: to be able to use alongside produce_and_store_eap_items. This is consistent with the rest of the test classes like uptime, profile functions, occurrences in the file.

Not too strong of an opinion though, lmk if you want me to just replace it haha

[wedamija]

You can just use store_eap_items though. I'm going to factor out all the store_profile_functions, store_trace_metrics, etc etc in some other pr. It's unnecessary to create new functions for each item type, the file name passed isn't used. I think AI is just replicating some pattern

[Abdkhan14]

You're right actually, getting rid of store_processing_errors

[Abdkhan14]

@wedamija this is done ^

[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.}

[cursor]

test_count expects 6 time buckets, but 7 exist

Medium Severity

The test_count test asserts against 6 time buckets (event_counts = [6, 0, 6, 3, 0, 3]), but test_zerofill with the identical time range and interval expects 7 buckets ([[{"count": 0}]] * 7). The test_top_events_group_by_error_type also implicitly acknowledges 7 buckets by slicing [0:6]. The event_counts list in test_count is missing a trailing zero-count entry for the 7th bucket, so the assertion will fail when the response contains 7 data points.

Additional Locations (1)

• tests/snuba/api/endpoints/test_organization_events_stats_processing_errors.py#L61-L74

 

[wedamija]

Thanks, lgtm!