May 1, 2002 at 8:48 am
Created merge publication with articles that are partioned based on field that matches to first 2 characters of remote servers name.
Using DCOM remote agent on low bandwidth connection.eg send all records for server 01(SQL) if field has 01 in it. Sql2000 SP2.
Normal push subscriptions never produce a single error ever.
Works fine except sporadically get error message - Merge process could not initialize the publisher - further details reveals
error - Another merge agent for the subscription(s) is running.
Looking in sysprocesses shows the process is incorrectly hanging around (normally goes after completion) and was last in the distribution database (presumably updating it with session details), so the error message above is correct as it will not allow 2 instances to run. Killing it and restarting the merge agent produces the same error over and over again. Leaving it for a few hours and it will often go away.
I do not think the TCPIP keep alive time is an issue as it is set to 5 mins.
Does anyone have any experience of this.
Thanks
Nigel Maneffa
May 3, 2002 at 5:27 am
The only item I find on this is http://support.microsoft.com/default.aspx?scid=kb;en-us;Q300180 but I don't think that is it. You may need to set Profiler up to trace the events for the merge publication to see if you can find a spot that doesn't execute or end properly. Use start and comelete events when looking and may want to catch to a to store for lookup. Make sure you catch the event data text and I beleive you can focus on SQLAgent to narrow down.
"Don't roll your eyes at me. I will tape them in place." (Teacher on Boston Public)
May 7, 2002 at 7:01 am
Thanks for the tip, I have only used profiler in its simple mode and simple filtering.
An update on the remote DCOM problem - the remote machine (which is on a site with no IT staff at all, hundreds of miles away) came up a NT error (NT4 SP6a) 2019 - The server was unable to allocate from the system nonpaged pool because the pool was empty.
This came up with machine idle (no users on site at all) and several hours later and 10 of the same errors later the machine went down and rebooted in the middle of the night. Since then (early days) the error has not appeared since. Until that point nothing else had been logged in the log. My (minimal) own findings of NT 2019 errors over the years have never proved anything conclusively, and a trip into MSDN unearthed little. Could it be that remote DCOM agents are an early warning of OS ailments !? Or maybe they are the cause of the problem !
Will keep posted. If happens again will use performance data log service to log conntinuous.
May 7, 2002 at 8:45 am
DCOM causes us some strange problems as well so I really cannot tell you for sure. We have even had objects we removed processes from still run those processes until we rebooted. There is no telling.
"Don't roll your eyes at me. I will tape them in place." (Teacher on Boston Public)
Viewing 4 posts - 1 through 3 (of 3 total)
You must be logged in to reply to this topic. Login to reply