Always rowlock in 4.0SP1

kriki

Someone has some experience or know more about "Always rowlock" and about possible problems in case we use it?
I found out that SQL is forced to always rowlock, avoiding deadlocks that are caused by page-locks.
But what about performance?
What about other consequences?

Thanx

kine

You need to think about that in this way: Always rowlock was enabled in older versions... :-) it means 3.60, 3.70, 4.00 w/o SPx were using the rowlock hinting.

You have performance problems when you are locking too much records - in some cases there can be thousends of locks in one transaction and it means BIG overhead. Disabling this can change it to just one table lock and better performance... it is more natural for MS SQL to work without rowlocking... :-)

kriki

Thanx kine.
I have a colleague who wanted to enable this (in the hope to get less deadlocks) and I told him to wait because I didn't know the consequences.
Now I know what it is (and most of the deadlocks I have already resolved in other ways).

freerain

Thk to both of you guys.

David.

jordi79

Hi All,

Just my 2 cents. I have tested on 2 clients. 1 client posting large transaction (> 1000 lines), and another client running a dataport to import into GenJnlLine (> 1000 lines).


With "always rowlock" disabled, I have having locking problems. While with "always rowlock" enabled, I am having no locking problems.

BTW... I am testing on NAV 5SP1 on SQL2005. And I tested on both clients with at least 2 journal lines apart from each other. This is to avoid the upper range and lower range locking.