Access DBs moved from Novell to Win2k Corruption problems after move

[shutech]

We just started a migrate from Novell 4.2 servers to Win2k servers. We have 2 databases backends in access 97. we have 2 clients written in access 97 also. First client a links to all tables in DB a which is 110 megs in size. Client B links to all tables in DB b which is 23 megs in size and also 5 tables from DB a. we also use access 2002 to run custom reports out of each database. This has been in place for 6 months on the novell server in its current form and for 4 years before that in the form of client and DBs in access 2.0 and report databases in access 97. Corruptions have been limmited, and I mean VERY limmited. (ie... power failed and a machines battery backup failed.)Monday I moved the databases off of the Novell servers and onto a win2k box. Brand new 1.2 PIII raid 5 and 1.2gig of mem Compaq server.

Since the move we have been getting database corruptions everyday. All have occured in DB a and all have been shutdown corruption of DB. I have used the 2 hot fixes talked about in the MS KB article Q303528. This is the same fix provided in a similar article but for a jet 3 engine. I don't have that KB ref at the time i am writing this. My next move is tomorrow to disable opportunistic locking.
I think I have narrowed the problem down to machines that are in Access client Bs all of a sudden rebooting. I went in and also unchecked from the startup and recovery button on the advanced tab in system property the automatically reboot option from system failure box. That was done on the 18th at noon and I did get the corruption on the 19th at 4:45 pm after a machine with the hotfix and the checkbox unchecked rebooted while in client b.

Any thoughts beyond disabling oplock on the server?

Like I said Novell fine WIN2k server not fine.


Michael

 

[thornmastr]

You are probably on the right track.

The primary reasons for corruptions to a back end on the server are a failing NIC on a users PC, or a hot reboot on the client machine while connected to the database. Either condition is almost guaranteed to cause corruption whether you are running a Microsoft or Novel network. However, if I remember correctly, Novel seems to be more robust in the face of some NIC failures. Perhaps you might want to replace the NICS on your client machines, at least on those you know have done reboots while in Access.
Robert Berman
Data Base consultant
Vulcan Software Services
thornmastr@yahoo.com

 

[fong]

Michael:

I experienced the same problem twice when an Access database was moved from a Novel server to a Win2000 server. When the database was moved back to the Novell server, few corruptions occurred.

After you disable oplock, does the database corrupt often?

Peter

 

[shutech]

Peter,

I have since corrected the problem by applying service pack 3 to all 2k workstations and 2k servers involved. Some of the information that I got off of microsofts site pointed to doing some manual config disable the oplocck and applying pre-sp3 hotfixes. I did those and still had problems. I could of missed a machine. But I did find out that the hotfixes were included in sp3 so I applied them. waited a month and then moved the databases back to win2k server. They have been up and running now without a single coruption since march 1, 2003.

the 2 databases are about 100M and 36M respectively and I had never had any coruption problems while on the Novell server.


To peg down what was done to fix it I don't know but the order of work went apply the 2 hotfixes. disable reboot on failure, disable oplock, and then apply sp3. Some items that might of had an affect but at the time was not troubleshooting the problem. and they were applied for other reasons were the latest jet DB engine came out and I applied that to all machines, To all serevers and those I had running SQL client I applied the latest at time MDAC 7.1.

The things to look for on microsofts web site are the KB articles that deal with optimizing JET DB engine. there are one for each version of the jet engine.

What it came down to was waiting a while, applying all the latest and then giving the move another shot. It was a hail mary play but my options were limmited. Since it was working on the Novell server and I was not being forced away from that server I couldn't dedicate enough time to go through a microsoft troubleshoot.

Hope this helps.


Michael

 

[Hap007]

Dear shutech,

I I also had a similar problem, but being stubborn, I tried to figure a layman's fix without any patches.

Here is what I came up with:

In Novell, the link to the shared backend was coded something like drive\directory\file ie.

Y:\Data\Database\MyBackend.mdb

This worked fine in Novell, but caused random corruption when a similar link to a Win 2K server was used.

When I Changed the link to:
\\ServerName\NewData\Database\MyBackend.mdb

for the new WIn 2k Server, the random corrution disappeared.

FYI:
Now that you are using Win 2K backend, don't you think that Novell was a more efficient method. Just proves that the best does not always win.

Hope This Helps,
Hap


Access Developer

http://www.fcs-software.com/fcsaccessdev.html

-
Specializing in Access based Add-on Solutions for the Developer

 

[shutech]

That is good to know too Hap!

Your right on the Novell thing. I was amazed yet not too all surprised that a complete MS solution (access on 2k machines and 2k servers) Failed while a hybred system (access on 2k machines and a novell file and print server) Worked. If I could of had integrated the Novell server without a great amount of time, money and difficulty. I would of left the server in place. But as a small business I needed 2k servers so I just did not want 2 different systems.

Thanks hap


Shutech