My database server performance is degrading over a number of weeks, see the charts attached.
I am running on a single node.
Previously I took a backup, wiped the database and imported it again to return to a clean system. At the same time I changed gc.ttlseconds=3600. However the performance is dropping off again in a similar way.
The database is backing Nakama as a game server and the application is write intensive. As the chart shows, disk reads become more intensive over time and the disk space taken is way above the game data stored, which is relatively static in size.
What are the possible causes of this behaviour and what steps can I take to rectify? Thanks!