October 22, 2008 at 12:36 am
Hi,
at first, Sorry for my bad English.
I have one SQL 2000 Server and a SQL 2005 Server. On the 2000 Server are 4 DB´s which replicated to the 2005 Server. The 2005 Server is the Distributor.
Three of this Replications work fine, but one isn´t syncron on some days. It take some hours to be syncron. Then is this db for somes Days syncron, and then suddenly not longer.
At the Sync-Status- Monitor are shown :"Transactions are replicated
completed" (in german, translated with google 😉 ) for hours, and suddenly it goes on.
I can´t find the reason.
Thanks
Andre
October 22, 2008 at 6:16 pm
Is the database which is not synchronizing properly getting lots of inserts, updates or deletes ?
What is the size of table being replicated ?
Does this happen at some perticular time in the day ?
If you are doing some huge batch upload operation on the published table then this might happen.
October 22, 2008 at 11:49 pm
Hi,
this databases is not the database with the most action on it. But some tables are very huge.
I mean, it happens at night, because i see the Problem in the Morning.
Huge batch uploads are not on this DB.
October 23, 2008 at 8:51 am
It sounds like you might have blocking. Next time you catch it happening I'd run a sp_who2 and see if you can observe a block in progress. Do you have a maintenance job running that's rebuilding indexes at the time and locking the table at the subscriber?
If not blocking, maybe your disks are just busy taking care of something else and transactions take longer than normal to apply. Maybe an ETL process is running and crunching a lot of numbers, or maybe an index is being rebuilt on some other table and eating resources.
In either case I'd begin by looking at what's going on at the subscriber at while the delay is occuring.
October 24, 2008 at 12:42 am
Hi,
there are not blocks or blockings on this database.
Yesterday, i builded the Replication on this database new, but today, the same problem.
October 24, 2008 at 7:02 am
What else is running on the subscriber at the time the delays occur? If there's no blocking your disks or CPU might just be busy and take longer to apply replicated transactions. If you know that it happens at a particular time set up a profiler session to capture what's happening on the server as well as a perfmon capture that you can analyze alongside the profiler results. You might also want to turn on verbose logging for your distribution agent to see if there are any timeouts or retries happening.
Viewing 6 posts - 1 through 5 (of 5 total)
You must be logged in to reply to this topic. Login to reply