There are two previous posts on this that seem to offer server-side solutions.  I have several DotNetNuke sites hosted on GoDaddy and the only problematic one is on their new "Grid" hosting plan.  Clearly there is a problem on the grids (probably a scaling issue) and GoDaddy and MS need to get that worked out.  It's not a static problem and comes and goes during the day, starting Wednesday 3/4/2009 after 3 full weeks of error-free testing.

It seems to me, though, that there could be some things going on in the (default) DotNetNuke setup that may be contributing to the problem.  I'd like to try to alter my configuration to ease the crunch on the number of calls stacked up in the buffer(s). 

That being said, I'm not skilled enough to know what that might be or where to start.  One thing I've noticed is that when the site (database) is up and responding, my error logs are absolutely "bloody" the time periods when erors are occurring.  That seems to me to be a possible problem contributor since once one error occurs the logging floodgate opens up and simply tons of detail is being recorded.  My event logs are now page-after -page of red.  I'd want to turn it back on at some point, but I'd rather keep the site up as much as I can until they fix the problem, or I can get everything moved w/o having to start over...

Any help for a poor shared hosting slave?