[BUG] Deadlock Count not resetting

[CADelete]

Component

Full Dashboard

Performance Monitor Version

2.4.1

SQL Server Version

SQL Server 2019

Windows Version

Windows Server 2019

Describe the Bug

After a deadlock has been received, the deadlock count and deadlock wait time (ms) does not reset and repeats the value in each collection. This is sporadic and doesnt happen every time but I have replicated it. Also by creating a new deadlock this sometimes resets it.

Steps to Reproduce

Open Dashboard
Create Deadlock
View Deadlock Count and see if it repeats

Expected Behavior

Deadlock count doesnt repeat and goes back to 0

Actual Behavior

Deadlock count continues at 1 for each collection

Error Messages / Log Output

No errors

Screenshots

Additional Context

SQL Server Agent is running, monitoring an Azure SQL VM with SQL Server 2019 installed. This has been an issue in the last 2 versions I have installed

[CADelete]

When lock wait is detected it then seems to affect the blocking events and deadlock count. The blocking events last for 1 hour then return to normal, the deadlock wait time continues

[erikdarlingdata]

Thanks for the detailed report and screenshots!

This was caused by a doomed transaction handling issue in the deadlock parsing pipeline. When sp_BlitzLock occasionally doomed the transaction inside process_deadlock_xml, the raw XML rows were never marked as processed. The next scheduled run would re-parse the same deadlock with a new collection_time, causing the analyzer to count it as a "new" deadlock each cycle — which is why the count persisted instead of resetting.

This was fixed in v2.5.0 via two changes:

• XACT_STATE check after third-party proc calls (#695) — detects and handles doomed transactions from sp_BlitzLock before they silently corrupt the pipeline state
• Doomed transaction error handling in delta framework (#756) — ensures ROLLBACK happens before error logging so errors aren't swallowed

v2.6 is releasing later today if you'd like to jump straight to that, otherwise upgrading to v2.5.0 will also resolve it.

[erikdarlingdata]

Closing — fixed in v2.5.0+.

[CADelete]

I've upgraded to v2.6.0 and the issue is still happening. I have installed the latest sp_blitzlock along with all of the other first responder scripts but no change. I did a clean install just to check if there was anything legacy based but the issue still occurs.

[erikdarlingdata]

Fixed in PR #820. The root cause: when no new deadlocks occurred for a database, the MERGE source CTE was empty, so no row was created for that database. The stale deadlock_count_delta from the last collection persisted because the delta calculation never ran again.

Fix: expanded the MERGE source to carry forward databases from the previous collection with zero counts. This ensures every previously-active database gets a zero row, so the delta properly resets to 0.