We're experiencing intermittent crashs and problems with our site. Our Moodle install has about 30,000 registered students and a bit over 2,000 course shells used primarily for "blended" instructional purposes. Many are just static, and the others are just starting to get active as we start up our school year.
We're running PHP Version 5.3.13, MySQL (I forget that version, but it's current as of June), Moodle 2.2.3+ (Build: 20120601) on Windows Server 2008 SP1 running IIS 7.5. Data (including file data) and web service are split between two boxes.
When things started goin' south, we'd get the following...
HTTP Error 500.0 - Internal Server Error
The FastCGI process has failed frequently recently. Try the request again in a while
...so we turned off FastCGI's auto fast recovery. We've had some problems getting PHP's error logs goin' too, but the error below seems the most likely crash-inducer...
[23-Aug-2012 09:17:23 UTC] Potential coding error - existing temptables found when disposing database. Must be dropped!
[23-Aug-2012 09:17:23 UTC] PHP Fatal error: Uncaught exception 'ddl_change_structure_exception' with message 'DDL sql execution error' in E:\www\lib\dml\moodle_database.php:400
#0 E:\www\lib\dml\mysqli_native_moodle_database.php(669): moodle_database->query_end(false)
#1 E:\www\lib\ddl\database_manager.php(88): mysqli_native_moodle_database->change_database_structure('DROP TEMPORARY ...')
#2 E:\www\lib\ddl\database_manager.php(75): database_manager->execute_sql('DROP TEMPORARY ...')
#3 E:\www\lib\ddl\database_manager.php(522): database_manager->execute_sql_arr(Array)
#4 E:\www\lib\dml\moodle_temptables.php(130): database_manager->drop_temp_table(Object(xmldb_table))
#5 E:\www\lib\dml\moodle_database.php(322): moodle_temptables->dispose()
#6 E:\www\lib\dml\mysqli_native_moodle_database.php(334): moodle_database->dispose()
#7 E:\www\lib\dml\moodle_database.php(140): mysqli_native_moodle_database->dispose()
#8 [internal function]: moodle_database->__destruct()
thrown in E:\www\lib\dml\moodle_database.php on line 400
...Does that actually look likely? If so, fixes? If not, any ideas where we might look?