May 16, 2002 at 3:39 am
Hi,
we are using a merge replication with a pull and a push subscription. Sometimes durring the replication we get the following message:
SQL Server Error: 11 General network error. Check your network documentation.
I checked the MSDN and found the following information:
CAUSE
The most common reason that this connection attempt failed is that this DSN or ODBC data source attempted to make a connection using the TCP/IP sockets Net-Library, which is Dbmssocn.dll. Because the SQL Server is not listening for incoming connections for TCP/IP sockets clients, the connection fails.
I checked the installation on the server and on the clients and it is okay. They are all listening on the same ports. The strange thing is, that the replication starts and durring the process receives data the connection is lost with the "General Network Error". Is it possible, that this message occurs durring gathering the data on the server with a mainprocessor usage of 100%? So, I guess a timeout happens. Is it possible to set the connection timeout for the sockets to a higher value?
Please help,
Tanks,
Thorsten
May 16, 2002 at 4:40 am
Not that I know of. Im currently seeing this on something Im working on as well (the network error part), looks like we have one segment with a lot of CRC failures, etc. Not much you can do except get the problem with the network fixed I think.
Andy
May 16, 2002 at 5:57 am
As you state this is merge replication and it could potentially be a timeout issue. I saw a statement in another thread about the rowguid column and that you should make it the clustered index to support faster merging. Might want to try and see what happens.
"Don't roll your eyes at me. I will tape them in place." (Teacher on Boston Public)
May 16, 2002 at 6:11 am
Thanks a lot,
I've allready tried to set the clustered index on the right column. It got better, but still we got the problem with the General-Network-Error.
It seems to happen when to many clients are replicating the same time. It's somehow like overloading the system.
Any idea? Will it get better with a bigger server or a few more distributers?
Thorsten
May 16, 2002 at 7:19 am
Before you try changing a lot of stuff, you really really want to get a network engineer to monitor the situation when the timeout is occurring. Just checking stats from the appropriate switches/routers involved may be enough to indicate if there is a problem or not.
Andy
May 16, 2002 at 7:39 am
I agree, you may also want to use Profiler to tarce both boxes to see if anything unusual is occurring.
"Don't roll your eyes at me. I will tape them in place." (Teacher on Boston Public)
May 16, 2002 at 7:41 am
Thanks,
I will try that.
Thorsten
Viewing 7 posts - 1 through 6 (of 6 total)
You must be logged in to reply to this topic. Login to reply