Skip to content

Health-parser reads say "No ... events were found" while the same answer counts 982 of them (ring-buffer truncation message) #6650

Closed

Description

@erikdarlingdata

Claude posting for Erik Darling.

What happens

When the system_health ring buffer is truncated, the health-parser reads use the truncation message from #6441. Its first clause says that no events of the type were found. The same answer reports hundreds of them in the window. The message should say that none matched the read's filter.

What I saw

v3.11.0 (1b6af51), get_health_parser_cpu_tasks with hours_back 24 on a busy SQL Server 2022 test instance. That instance's system_health ring buffer is at SQL Server's 4 MB cap.

"status": "unavailable",
"message": "No sp_server_diagnostics_component_result events were found for <server> in the last 24 hour(s), and this is NOT an all-clear: the newest system_health collection run reported that the Extended Events ring buffer this collector reads is truncated (...): 218 event(s) are hidden from the read and the newest visible event is 1 h 54 min old, so newer events are not being collected. ...",
"source_observed": true,
"last_captured_of_type_at": "2026-10-23T13:22:47.5435790",
"events_in_window": 982

On 3.9.0, before #6441, the same read said: "1035 sp_server_diagnostics_component_result event(s) were captured for in the last 24 hour(s) and none was a QUERY_PROCESSING component result in the WARNING state with at least 10 pending tasks." That text named the count and the filter. The new text drops both.

The same message appears on get_health_parser_io_issues, get_health_parser_memory_conditions, get_health_parser_memory_broker, get_health_parser_memory_node_oom and get_health_parser_scheduler_issues. I checked the event count only for get_health_parser_cpu_tasks.

Where

PerformanceMonitor.Common/RingBufferTruncationNote.cs:113 HealthParserMessage. Its summary says it is used "when a category came back empty". Here, empty means no event matched the category's filter, not that no event of the type was stored.

Suggested fix

Pass the event count and the filter description into HealthParserMessage. Then keep both clauses, for example: "982 sp_server_diagnostics_component_result events were read for in the last 24 hours, and none was a QUERY_PROCESSING result in the WARNING state with at least 10 pending tasks. This is not an all-clear: ... the newest visible event is 1 h 54 min old." Use "No ... events were found" only when the count is 0.

How to verify

A test with events_in_window greater than 0 and a truncation note must not produce "No ... events were found".

Activity

  1. added
    bugSomething isn't working
    client-siteOwned by the client-site agents (other laptop). Local sessions never pick these up.
    on Oct 23, 2026
  2. erikdarlingdata commented on Oct 24, 2026

    @erikdarlingdata
    OwnerAuthor

    Claude posting for Erik Darling. Fixed on dev by #6664 (b9411a0). Ships in the next release.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't workingclient-siteOwned by the client-site agents (other laptop). Local sessions never pick these up.owner:pm-worker-coordinatorLaptop coordinator session

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions