Yesterday morning, we found and fixed an error in the way the site handled web crawlers, which was causing threads to block. We also did additional tuning of cache sizes. These changes resulted in a significant increase in stability and performance: the site remained up for 22 hours with much better performance than we had seen.
We continue to monitor the site and are investigating the cause of this morning's restarts, but it appears that we have turned a corner, given site visitors are also reporting improved performance. We continue to investigate performance, as we believe page load times can be further improved. The main area of work, as I wrote earlier, is to cluster the application server, which will allow us to increase overall cache size and take advantage of load balancing. This will result in faster performance and increased flight time of the application.
Good job! Please keep up a good work. There are still few annoying items, but forum became much more usable!