I have replication between server A,B,C.
A is publication server, B, C is subscription server.
Windows NT .SQLServer 6.5 spa5.
Network : A to B 10M Lan; A to C 100M Lan
B and C has same database and subscribs same publication,same article.Today
I found there is something wrong with the data of server B In fact , B
server didn't successfuly executive the jobs from 2270298
to 2270313 at 1:19PM; But C server had successfuly executived it.
Their is no error with distribution task (B and C) and no error happened on
Server B. I sure that the distribution task had sent out the jobs to the
distribution server B because I found the status of jobs from 2270298 to
2270313 about subscription B is 0 in MSsubscriber_status table.
Server B is heavy than server C (both application and network).
Data changes very fast from 1:00PM to 3:00PM (about
Immediate-updating). One article's table have no relation with
other article's table.
Is it possible that distribution task executived the transaction but
B is heavy that time, so B didn't successfully checkpoint them (write
the changes to disk), then the transaction lost? The recovery interval
is set to 5 minute.
I have file about distribution tables.
Can you give me some suggestion?
Your help will be great appreciated.