Mark, the issue with timer mode is that it might not have portal context and some jobs might fail. request mode is suggested setting and if your keep alive service calls a real page, it will trigger the scheduler as well.

---

Cheers from Germany,
Sebastian Leupold

   

Speed up your DNN Websites with TurboDNN

I just wanted to give an update, I have had no occurrences of the problem after switching the scheduler to Timer Mode (about 4 hours ago now).  It was coming very consistently before this change, as I said always in-sync with a request I made to a portal.  I was getting 25+ of the messages at once (or all within a 60-second period) so this may not be the same thing you are experiencing.  My scheduler log showed the Purge Module Cache and Messaging Dispatch jobs taking the same 60 seconds to complete, when normally they took fractions of a second.

To ensure I had my best foot forward for performance I also took most of the steps outlined in this blog entry from Mitch Sellers:
http://www.mitchelsellers.com/blogs/a...

We already had most of the settings set to match these but this filled in a few gaps.  The only thing that I feel is relevant to this problem, aside from the scheduler mode, is regular purging the EventLog table.  We have our own system so I can schedule this in SQL Server Agent.  In the event that type of automatic operation isn't possible it seems limiting the events going into the log is the best option; I turned off all of the "success" type logging and that has already kept it vastly cleaner.

Sebastian, I don't understand what you mean by portal context and how that would impact the scheduler.  If we were talking security contexts I'd follow.  Fortunately looking at the scheduler logs I don't seem to have any negative impact of the mode change at this time, but I don't discount the concern and would like to understand what you mean.