Maintenance window today — 9PM-9.20PM GMT+1

By Edouard on May 4, 2010

If you use Web Translate It extensively you might have noticed a few error pages yesterday, when loading pages or saving strings.

The reason of these errors is that the database server is sometimes overloaded and crashes at peak time.

I take this problem very seriously, and in fact I think most of the issues are now fixed. I tuned the database server, pushed some important optimisations that will make the service much more reliable than yesterday. I will carry on working on optimising the service today.

Tonight I will also upgrade Web Translate It’s hardware, which will require to reboot the server. Rebooting the server should take less than 5 minutes, and the service should be completely operational a few minutes after that. Therefore I reserve a window of 20 minutes downtime tonight from 9PM to 9.20PM GMT+1.

I apologise for the short notice and for the inconvenience this downtime will cause, but this downtime will be very short and make the service a lot more reliable.

EDIT: The maintenance went well. The service was down for about 4 minutes, according to Pingdom, and it took another 5 minutes to restart the workers and search engine before the service was 100%.

I upgraded Web Translate It’s slice memory from 1.2GB to 1.5GB, and I increased the database cache. Some indexes were corrupted due to the crashing, so I also rebuilt the indexes.