[BUG] Issue with PerformanceMonitor_BlockedProcess and PerformanceMonitor_Deadlock Extended Events

[gotqn]

Component

Lite

Performance Monitor Version

2.11.0

SQL Server Version

Microsoft SQL Server 2022 (RTM-CU19) (KB5054531) - 16.0.4195.2 (X64)

Windows Version

Windows Server 2019 Standard

Describe the Bug

In the Lite edition, the PerformanceMonitor_BlockedProcess and PerformanceMonitor_Deadlock Extended Events sessions are created only on the "open the server tab" code path. The recurring background collection loop never (re)creates them. As a result, if that first creation attempt doesn't succeed (connection not fully established yet, or insufficient rights at that moment), the sessions are never created and never retried automatically — the blocked_process_report and deadlocks collectors then silently read a non‑existent ring buffer and report success, so there's no indication anything is wrong.

Steps to Reproduce

1. Add a server and let background collection run without opening the server tab (e.g., app minimized to tray after restart). → XE sessions are never created.

or

2. On the very first tab‑open, have the create attempt fail (connection not yet established, or the login temporarily lacks ALTER ANY EVENT SESSION). → it fails once, and no later background cycle retries it.

3. Observe: blocked_process_report / deadlocks collectors show OK, but sys.server_event_sessions has no PerformanceMonitor_* sessions and the Blocking/Deadlock data + Blocking alert stay empty indefinitely.

4. Manually re‑open (double‑click) the server tab with a live connection and adequate rights → sessions are created and everything starts working. (Confirmed.)

Expected Behavior

XE sessions should be ensured as part of the recurring collection lifecycle (or retried), so that:

• servers monitored without an open tab still get their sessions, and
• a failed first attempt self‑heals on a later collection cycle instead of requiring a manual tab re‑open.

Actual Behavior

Sessions are created only on RunAllCollectorsForServerAsync (tab‑open). The background loop never ensures or retries; a failed first attempt leaves blocking/deadlock collection permanently broken until a manual tab re‑open, with no UI signal.

Error Messages / Log Output

Screenshots

No response

Additional Context

Suggested fix

• Call EnsureBlockedProcessXeSessionAsync / EnsureDeadlockXeSessionAsync from the background collection path (e.g., once per server per cycle, or guarded so it only re‑attempts when the session is missing), not only from RunAllCollectorsForServerAsync.
• Add a null/ready guard at RemoteCollectorService.cs:349 so the ensure step is skipped cleanly (and retried next cycle) when GetConnectionStatus isn't populated yet, instead of bailing out.
• Surface the ensure failure beyond a LogWarning — e.g., mark the blocked_process_report/deadlocks collector health as degraded so the UI doesn't show "OK" while capture is non‑functional.

[erikdarlingdata]

Fixed in #1089 — XE session creation (PerformanceMonitor_BlockedProcess / PerformanceMonitor_Deadlock) was moved into the shared RunCollectorAsync path, so the background collection loop now ensures/retries the sessions every cycle, not just on tab-open. A server monitored without an open tab gets its sessions, and a failed first attempt self-heals on a later cycle.

The "silent success" hole is also closed: a failed ensure now throws and is classified as an error/permissions failure, so the collector can no longer report SUCCESS while reading a non-existent ring buffer.

On dev, ships next release. Closing.