I still don't have a fix for the bug that occasionally causes the application to fail, but I do have a workaround.  I will be running multiple copies of the application so that if any one instance fails, the other copies will take up the load until the failed instance can be restarted - which should happen within a matter of seconds.

I'll be putting this in place tomorrow; unfortunately the change itself requires a brief outage, so the system will be unavailable for up to 5 minutes from Friday 13th April 2:00 GMT.

Right now - you may have seen a brief hiccup - I have put a new and more reliable automatic restart script in place.

This should give us a more stable system while I track down the problem.

Update: The change is in place - five minutes early.  We now have load balancing and automatic failover between three copies of the application.

Posted by: Pixy Misa at 10:19 AM | No Comments | Add Comment
Post contains 158 words, total size 1 kb.