March 24, 2003 at 10:38 pm
hi all.
I have transactional and merge replication happening in one of my production server. I restart my agent every day in the morning in the office. After restarting the merge agent failed and gave the ffollowing eror saying "The gaent is suspect no rewsponse within last 10 minutes" I had to restart the SQL Server service and then its tarted. Does any1 knows what is the problem, I also searched on BOL, for that error, but then there is nothing there in that, and when i went to the mocrosft site it says go to the INF article 190388, but when u click on that libk it says u r looking a page which is not available. Does any1 knows a problem about this.
Thnks in advance
March 25, 2003 at 5:40 am
I have that error usually because of out slow WAN connection. Is not a big problem, maybe if you try stating the agent again later. This was the distribution agent right?
March 25, 2003 at 5:45 am
No sir, the merge agent was failing and it was saying that The agent is susupect, No action to be done in the last 10- minutes, I had to restart my SQL Service for fixing the problem, thank god it was morning and i was in office and connected to the server remotely and i was monitoring something else, or else, i would have gone, and fired from my boss
March 25, 2003 at 6:31 am
I get this occassionally and believe it's due to communications failure. Most subscribers are using dialup access. All I do is start the merge agent again and it's fine. You shoulnd't have to restart the server.
Ten minutes is the default timeout setting, and is enforced by a job on the publisher that checks for "dead" replication agents. You can modify this job to notify you when that happens so you can take immediate action.
Jay Madren
Jay Madren
March 25, 2003 at 7:17 am
Yes it's a timeout problem, a connection problem. Re start the agent. Agree to set the job to notify you.
March 25, 2003 at 9:36 pm
ya notifying is fine, but what if i am not there i n the office, or i am at a place where i cant connect to the server, then i am really gone.
March 25, 2003 at 10:11 pm
You said in your first post that you "restart my agent every day in the morning in the office", but now you're talking about not being there. So, are you starting the agent manually or is it on an automatic schedule?
I see two options: If it is scheduled to run automatically, schedule it to run more often so if one fails the next will pick it up; If you're running it manually and then leaving before it finishes, then modify the agent's job to repeat the step if it fails.
Jay Madren
Jay Madren
March 25, 2003 at 10:35 pm
Hey , can u tell me how to restart agent automatically, if u can tell me that , just gr8.
Thnks in advance
March 26, 2003 at 7:54 am
I Guess you could create a script that executes the cp_cmdshell 'NET Stop SQLServerAgent' command. But you coudn't schedulle it with a job, so you should use a batch script that executes osql and schedulle it from windows.
March 26, 2003 at 8:27 am
It's not the SQL Server Agent that needs restarting. It's the merge replication agent, which is started by a job. To have the merge agent restart (or rerun) automatically, find the job either on the subscriber (for pull) or the publisher (for push) that runs the merge agent for this subscription. In that job, create another step by copying the Type and Command from the step that runs the agent. Then set the original step to run this new step on failure. Note that passwords in the command may be encrypted, so you'll have to replace them with the actual password in the new step - copy & pasting the encrypted passwords won't work.
Jay Madren
Jay Madren
March 27, 2003 at 3:21 am
Try this
Under Replication Monitor/Replication Alerts - configure the replication retry agent - response tab - check 'execute job' & select the merge agent job from the drop down menu. Done this here & it works a treat - almost never have to intervene!
March 27, 2003 at 8:16 am
We used to have this problem as well. Another idea ... find the job that runs the merge agent (as jmadren describes), go to the Schedules tab, click New Schedule, and add a recurring schedule that runs as often as you like (once an hour, once a day, whatever). This will restart the task on the next schedule cycle, whenever it fails.
Something to keep in mind with both this and jmadren's solutions is that anytime you tear down and recreate the publication, you have to make these manual changes the task again.
Randy Rabin
March 27, 2003 at 9:44 pm
Thnks guys for al ur help
I have sloved the problem by making the agent run in a cyclic fashion
and i will also try putting up the replication alert
Regards
Jay
Viewing 13 posts - 1 through 12 (of 12 total)
You must be logged in to reply to this topic. Login to reply