New backup & restart process, minimized downtime

I have revamped the backup & restart process. In the past, the server would shutdown, make all the backups and restart. This would take about 10 minutes (its about 2GB of data). The backup includes the plugins, the server worlds, the website, the logfiles and all backend code written by me that runs all kind of stuff.

From today on, the process is different. The server now stops saving the world changes while running, can thus back up the world file & plugins w/o shutting down. During that, the system backs up the world files. When finished (after 10 minutes) it will save the not saved changes to the world file, resumes saving in general and then backs up all the rest. Then, the server will shut down, backup the logfiles only and then immediately restart. This shortens the downtime of the server to less than a minute.

The advantage is that the server downtime is influencing our ranking on the server list. We are listed there as having an uptime between 99.3% and 98.1%. Every percent higher is a higher ranking chance in the list. Also, it allows people to relogin w/o waiting for 10 minutes after a shutdown. The restarts are still needed to reclaim memory for the server that some plugins will not set free and therefore slowly slowing the server down over time.

The disadvantage is that during 10 minutes, in case the server would crash right then, the world changes done in those 10 minutes would be lost. Which is a really neglectable issue, specially since the downtime happens during the time when there are the fewest people on the server.

2 thoughts on “New backup & restart process, minimized downtime

Comments are closed.