Looks like we have a problem with the scheduler on antville.org. A grep from the access log:

[2003/02/06 13:48] antville:internal:scheduler 333 marked, 0 inserted, 333 updated, 0 deleted in 898358 millis [2003/02/06 13:56] antville:internal:scheduler 248 marked, 0 inserted, 248 updated, 0 deleted in 455056 millis [2003/02/06 14:04] antville:internal:scheduler 93 marked, 0 inserted, 93 updated, 0 deleted in 473335 millis

Possible problems are DB write performance or concurrent write locks. I'm investigating...

OK, as far as I can see, one probelmatic section is this in global writeReadLog():

      var site = root.get(el.site);
      if (!site)
         continue;
      var story = site.allstories.get(String(el.story));
      if (!story)
         continue;
      story.reads = el.reads;

What happens here is that for each story being updated, the site and its full story index needs to be loaded. I have checked in a mod to root/type.properties that allows stories to be loaded via root/storiesByID without the whole story index being loaded (yes, specifying an accessname does have that effect - I know that's an insider trick ;-).

I'll change writeReadLog() to use that to access the stories:

      var story = root.storiesByID.get(String(el.story));
      if (!story)
         continue;
      story.reads = el.reads;

Is that OK with everybody?