matthew.green 36969 (6/12/2015)

---

thanks - i have opted for slightly simpiler way of doing it:

using the app lock sp - basically queues up the calls to proc so only 1 user...

You have concurrency issues because your code reads the recordid before writing to the table and locking it, so multiple connections could read the same recordid under the default transaction...

Sad news. Great actor.

Wild guess is that the index was not covering and you got the classic lookup deadlock SQL Server is (in)famous for.

SNAPSHOT isolation is usually the cure for this desease.

I agree that ARITHABORT is a false fix, but hardcoding a hint is the last resort in my opinion.

I would rather focus on understanding which cardinality estimation goes wrong when...

I had to replace my windshield twice last year. Twice in two months to be precise. Not funny.

A small rock in the motorway is enough to break it and even...

I do: she fixed my mistake 🙂

Good luck!

Try this:

EXEC adasdb.master.sys.sp_executesql 'RESTORE DATABASE ETLNewTest FROM DISK = ''g:\MSSQL11.MSSQLSERVER\MSSQL\ETL_663.bak''';

Any difference?

Those stats are created automatically by SQL Server. If you want to replace them with custom statistics, you can do it.

Glad you sorted it out

ARITHABORT is a client setting that can be changed at any moment.

If you change your SSMS connection setting and clear the cache for that plan ( DBCC FREEPROCCACHE(plan_handle) ) does...

If you correlate the plan from...

At a glance, looks like the previous installation of service pack 3 failed to update the binaries, which is quite weird.

Adding more nodes to a SQL 2005 cluster has always...

The SQL Agent job is probably running a different execution plan compared to the one you get when you run the query in SSMS.

The reason might be one among:

* stale...