Skip to content

Fix sp_BlitzLock parallel deadlock filtering regression #3887

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
BrentOzar merged 2 commits into from
Apr 6, 2026
Merged

Fix sp_BlitzLock parallel deadlock filtering regression #3887

Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
3e0c5aa
Fix sp_BlitzLock qn column to restore 0-based numbering
BrentOzar Apr 6, 2026
8a1b8bf
Fix deadlock_group query numbering label for 0-based qn
BrentOzar Apr 6, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions sp_BlitzLock.sql
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3618,7 +3618,7 @@ BEGIN
en =
DENSE_RANK() OVER (ORDER BY dp.event_date),
qn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date),
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date) - 1,
dn =
Comment on lines 3618 to 3622
Copy link

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

qn is now 0-based (ROW_NUMBER() - 1), but downstream formatting later builds deadlock_group with CASE WHEN d.qn = 0 THEN '1' ELSE CONVERT(..., d.qn) END, which will label both qn=0 and qn=1 as "Query #1". Update the consumer to display d.qn + 1 (or otherwise adjust) so query numbering remains correct, especially for parallel deadlocks where only qn<2 rows are kept.

Copilot uses AI. Check for mistakes.
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.event_date),
Comment on lines +3621 to 3623
Copy link

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The ROW_NUMBER() uses ORDER BY dp.event_date inside PARTITION BY dp.event_date, which is non-deterministic because all rows in the partition have the same ordering key. Since qn is used to filter parallel deadlocks to the first 2 rows, consider ordering by a stable discriminator (e.g., dp.id and/or dp.spid) to ensure consistent row selection.

Suggested change
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date) - 1,
dn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.event_date),
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.id, dp.spid) - 1,
dn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.spid, dp.id),

Copilot uses AI. Check for mistakes.
dp.is_victim,
Expand Down Expand Up @@ -3703,7 +3703,7 @@ BEGIN
en =
DENSE_RANK() OVER (ORDER BY dp.event_date),
qn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date),
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date) - 1,
Copy link

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same as earlier: with qn now 0-based, the later deadlock_group label logic (WHEN d.qn = 0 THEN '1' ELSE CONVERT(..., d.qn)) will mislabel qn=1 as "Query #1". Adjust the label computation to use d.qn + 1 (or equivalent) so output query numbering stays correct.

Suggested change
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date) - 1,
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date),

Copilot uses AI. Check for mistakes.
dn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.event_date),
Comment on lines +3706 to 3708
Copy link

Copilot AI Apr 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Same non-determinism here: ROW_NUMBER() orders by dp.event_date while partitioning by dp.event_date, so qn assignment can change between runs/plans. Because later logic filters qn < 2 for parallel deadlocks, order by a stable column (e.g., dp.id, dp.spid) to make the filter deterministic.

Suggested change
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.event_date) - 1,
dn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.event_date),
ROW_NUMBER() OVER (PARTITION BY dp.event_date ORDER BY dp.id, dp.spid) - 1,
dn =
ROW_NUMBER() OVER (PARTITION BY dp.event_date, dp.id ORDER BY dp.spid),

Copilot uses AI. Check for mistakes.
is_victim = 1,
Expand Down