October 24, 2002 at 4:29 am
My general network errors are back!
Very often (once a day) my transactional replication fails after losing communication with the subscriber usually mid update and then it cannot reconnect and eventually gives up trying. As far as I am aware there is no corresponding general network failure that anyone else has noticed and as its an always on leased line there should not be any dropouts.
When I look at the agent history it tells me that there are problems with the ODBC connection, these are either connection timed out, communication link failure, then I get a report that the subscriber sql server does not exist or access denied, then lots of timeout messages again then after the 10 retry attempts it stops;
1. Why is it using ODBC
2. Why is it timing out
3. Is there somewhere I can alter the timeout properties.
As an aside is there a recommended process for completely erasing all replication history, SPs, etc etc from both the subscriber and publisher and starting again on a clean replication slate..?
Thanks for any help you can give
Rolf
October 24, 2002 at 6:44 pm
Dont know if I can answer all...or even any! You can change the timeout of the distribution agent, easiest way is to change/add a param to the job. You can also turn on logging to a disk file so that you can see what happened. Makes sense to try to figure why first, otherwise you may just hide the problem. Could be hitting a deadlock, subscriber server is overloaded, etc, etc. Leased line or not, you could have problems along the way. Have someone check the router/switch involved for bad packets, crc errors, that type of thing. No idea on the ODBC - suspect it was already built and they didnt want to change. Good enough for what it does I think.
Andy
Viewing 2 posts - 1 through 1 (of 1 total)
You must be logged in to reply to this topic. Login to reply