System down Scenario,

jannestig

We recently had a scenario which effectively shutdown a 4.0 SQL client with 50 users where every user who used a posting routine was effectively blocked in a waterfalling affect ( the use fullblown navision manafacturing warehouse etc) no matter which routine it was the end client would freeze and block other users although the process would actualy be completed.

Closing the client would resolve this (they all use citrix).
The issue that changed on the server was that someone had installed a stand alone database on the same SQL server and upon its removal the process and block dissappeared.

I can only assume somehow this standalone DB affected the way navision clients were recieving "terminate process" traffic.

Has anyone encountered this or can hypotheis upon the reason for this?

jlandeen

That really sounds like a strange scenario. I've never seen issues in the past with multiple databases running on the same SQL server. Was the standalone version of the DB a clean cronus database or was it a copy of your production database?

Was the license stored in the database or in the server?

kine

In some older versions of client there was bug, when during the Message or Error is on the screen the transaction is still blocked. May be that the client was in this state and the Citrix disconnected the session because the Idle time.

jannestig

Assuming Kines answer is not the cause the client responded with the following to recreate the scenario.

"Quite simple, create a new SQL database using default options. Restrict the access to be single user. You can do this now very easily or just do this on an existing test db, etc.

This is required for BackUp Exec to restore a database copy into, which was not performed today due to issues as you know."

On recalling exact details the SQL 2000 server had 4 DBS 1 production 2 testing (copies of production) on there the standalone DB had the same name as one of the tests DBs,

Also of not i RDPed into the server at the same time and posted with success, we had someone on a laptop with client install run a post as well so citrix doesn't appear to be the answer

kine

The best how to check what is the problem is to during the block check locks and blocks and find out the block chaing head.

jannestig

abit of a problem with that its is its not until 4,01 you have the options

"Database options have been added that allows SQL Server to determine lock granularity and to set the lock timeout."

It is not a true block as the original posting process was not completing all subsequent posting routines through all modules were block as well this was occuring even when only 1 user was logged in during intial debug, like the CIO i am happy enough it was solved but i hate not having a clear understanding of cause and resolution ](*,)