I wondered if anyone had seen this? I do have some workarounds but these involve more code changes than are ideal. Problem is specific to compatibility level 170 from what I can see.

With a trigger like this:

CREATE OR ALTER TRIGGER CH_REPRO 
ON CH 
INSTEAD OF UPDATE
AS 
BEGIN
 DECLARE @cirObj INT = 101;

 SELECT U.cir
 FROM inserted AS I
 LEFT JOIN TC AS U ON U.cir = I.itemno
 AND I.objectno = @cirObj;
END;

Runing any update such as

UPDATE CH 
SET bearer = bearer;

results in an error:

Msg 596, Level 21, State 1, Line 24
Cannot continue the execution because the session is in the kill state.

Msg 0, Level 20, State 0, Line 24
A severe error occurred on the current command. The results, if any, should be discarded.

The view or table CH seems irrelevant to the issue, but in this case was defined as

DROP TABLE IF EXISTS CH;

CREATE TABLE CH 
(
 bcport int,
 channel int,
 bearer int,
 objectno int,
 itemno int
);

INSERT INTO CH (bcport, channel, bearer, objectno, itemno) VALUES(31,1,999,null,null);

Table TC is a fairly large table, rather too wide (200+ columns) typically 100k+ rows. It does have multiple indexes on it, including a clustered index.

Changing the trigger to hard code the integer in place of variable @cirObj removes the error.

Changing table TC to be an expression such as

(SELECT * FROM TC)

or a CTE removes the error also.

Changing compatibility level to be 160 or lower removes the error.

I appreciate that selecting from a trigger isn't normal, the more complex original of this was doing an update with the same result.

I've reproduced this in Microsoft SQL Server 2025 CU5. Checkdb and similar report no issues. Table TC can be read and updated as normal outside of this use, and issue is on multiple databases I've tried.

Is anyone aware of a bug report or workaround for this that doesn't involve recoding the relevant triggers?

Stackoverflow

Message History

Delivery option(s)