May 7, 2010 at 1:21 am
Hi Experts,
I have a Merge push replication setup from US to a Korean server.Replication looks fine but intermittently I observe few errors in eventlogs and replication monitor: DESCRIPTION: Replication-Replication Merge Subsystem: agent
US-GlobalBOB_Prod-GlobalBob_Prod- Korea-7 failed. The merge process was unable to
access row metadata at the 'Subscriber'. When troubleshooting,
restart the synchronization with verbose history logging and specify
an output file to write to, or use SQL Profiler to determine the
source of the failure.
I ran a profiler and noticed that the replication thread is live and all of sudden it shows AUDIT LOGOUT for it.It again catches up by itself.I try to run a query remotely to that server.It sometimes runs well but then often I get an Transport level error whcih happens when our thread gets killed.
I know there is something which kills the thread from Publisher to subcriber servers.But how to find it.It happens intermittently not too often.Please share your inputs
May 10, 2010 at 1:31 am
Can somebody please throw some light on this(even if the exact fix is not known)
May 10, 2010 at 1:38 am
I try to run a query remotely to that server.It sometimes runs well but then often I get an Transport level error
To me this is more telling that the replication issue.
It just sounds like an unreliable network connection somewhere in the chain.
How to diagnose that to find the weak link i dont know, speak to you networking dept.
I know there is something which kills the thread from Publisher to subcriber servers
Are you talking about the SQLServer product , or something other task ?
May 10, 2010 at 10:56 am
Dave Ballantyne (5/10/2010)
I try to run a query remotely to that server.It sometimes runs well but then often I get an Transport level error
To me this is more telling that the replication issue.
It just sounds like an unreliable network connection somewhere in the chain.
How to diagnose that to find the weak link i dont know, speak to you networking dept.
I know there is something which kills the thread from Publisher to subcriber servers
Are you talking about the SQLServer product , or something other task ?
I checked my profiler trace.The replication thread shows 'AUDIT LOGOUT' in profiler trace.
A ping for 6 hours shows no issue between the server.The SQL Serevr is set with all default properties.
May 11, 2010 at 8:34 am
The error "Transport Level error" usually happens when the connection to the DB is disrupted. It could either be the network layer or your DB is disconnecting your session for some reason. You will have to check the error logs of SQL Server on both system to confirm what is going on. Please check the error log and let us know if you see any restart of SQL Server or some disconnects.
-Roy
May 11, 2010 at 10:39 am
Roy Ernest (5/11/2010)
The error "Transport Level error" usually happens when the connection to the DB is disrupted. It could either be the network layer or your DB is disconnecting your session for some reason. You will have to check the error logs of SQL Server on both system to confirm what is going on. Please check the error log and let us know if you see any restart of SQL Server or some disconnects.
You are correct.This error occurs due to interupptions.But there is nothing informative in SQL log.The auto-close option is also false.Trace just shows 'AUDIT-LOGOUT' and no reason is specified.
May 11, 2010 at 10:42 am
Today I had a look at the Default Agent Profile properties for my merge replication.There is a parameter 'Login Timeout' which is set to 15 seconds by default.Could this be an issue?? Is it possible that if the thread gets logged out after 15 seconds.The ping data between two servers over a period of 6 hours looks fine.Can there be any other reason? Please help me.
May 11, 2010 at 11:23 am
I have set up replication between where I am (Caribbean Isles) to Asia. The one thing that we had was VPN and a pretty good connectivity. We never faced this issue. The parameter you are talking about is for login. If SQL Server is not able to connect within 15 sec, it will just timeout. That is not the issue.
maybe the packet is being dropped in between causing this. You probably will have to get your network admins help on this.
-Roy
May 14, 2010 at 1:08 am
Roy Ernest (5/11/2010)
I have set up replication between where I am (Caribbean Isles) to Asia. The one thing that we had was VPN and a pretty good connectivity. We never faced this issue. The parameter you are talking about is for login. If SQL Server is not able to connect within 15 sec, it will just timeout. That is not the issue.maybe the packet is being dropped in between causing this. You probably will have to get your network admins help on this.
thanks for replying(during the T20 world cup in Carribean:-P )
As you mentioned that there are chances of few packets being dropped.But when i ping both the servers,the connectivity is continuos.I do not have much networking skills.But doesn't it look contrary?
May 17, 2010 at 7:18 am
When you are pinging, is it continuous? Not even one packet dropped? What is the latency between the two servers?
I think you will need to involve the Network admins in this issue since it looks more like a network issue. Maybe there is a firewall that is wrongly configured. Even if there is an audit logout, it should login automatically when the server tries to connect.
-Roy
May 17, 2010 at 9:43 am
Roy Ernest (5/17/2010)
When you are pinging, is it continuous? Not even one packet dropped? What is the latency between the two servers?I think you will need to involve the Network admins in this issue since it looks more like a network issue. Maybe there is a firewall that is wrongly configured. Even if there is an audit logout, it should login automatically when the server tries to connect.
I have asked the windows and Network teams to find out if there is any latency issue between two servers.I will keep you updated with the progress.
Thanks Roy!!!
Viewing 11 posts - 1 through 10 (of 10 total)
You must be logged in to reply to this topic. Login to reply