Amazon RDS “Freeable Memory” decreasing rapidly
I have noticed recently that freeable memory on my Amazon RDS (db.m4.large instance) is decreasing rapidly. Over the past few weeks, it has come down rapidly from almost 3700 MB to 1000 MB. There hasn't been any issues with the applications though. My CPU utilization is also consistently low - somewhere around 7 % all the time.
Is there something I can do to clear the memory? Worried if it goes at this rate, it may run out of memory in the coming days.
A downward trend of minimum FreeabkeMemory isn't alarming. It means your database process is requesting memory from the OS. As for why the database is dining that, you'll need to refer to your engine's vendor documentation for measuring internal memory structures and relate them to you workload. JOINs and ORDER BY queues are always frequent memory consumers as they'll need to allocate buffers for work tables. Depending on your engine or schema, that could also lead to temporary or long term disk usage, evidenced by minimum FreeStorageSpace drops and spikes in maximum WriteIOPs. Or even a sudden change in SwapUsage.
If you restart the instance, you'll see all of the memory allocated to the database process freed and the minimum FreeableMemory will spike. Then once again, the metric will slowly trend downward.
Ultimately, you want the database to allocate memory from the OS. It's sudden changes in trends that are worth investigating and the application (database) level.
And if you want more OS metrics you can always enable Enhanced Monitoring: http://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_Monitoring.OS.html