The Question and Answer Board

[Bob Zumbrunnen]

Should be. I'm not, though. <g>

When I finally was able to contact a human (James, who's been our reliable "switch-monkey" and more all along, but currently inaccessible because of their phone system), he told me SI's db server was sitting there powered down as if it'd just been taken out of the box.

Not good. Dave and I are pretty sure iHub's db server did exactly this same thing a couple of weeks ago. And when it happened here, about 1000 messages were lost.

SI didn't drop a single message. We need to do a comparison to see what's different in iHub's configuration because SI showed an event log entry that it had "rolled forward" 4333 transactions when it fired back up. Leading me to believe that write-caching is happening on that machine like it apparently is on iHub's (and it's extremely aggressive caching, apparently going upwards of an hour without committing data to disk) but SI's db is doing more aggressive logging and its log entries are committed to disk immediately.