Is rsync really a culprit for causing low memory on our VM?

Question

In a VM of 16GB RAM, we are running rsync as a cron job(for every 10 minutes) on our production to sync GBs of folder from AWS EFS to local storage. After few days of running we found that the VM is running low on memory and we found that size of our buffer/cache is more than 4GB and using vmtouch we confirmed that entire folder we are syncing is getting cached.

Since we are running this on production we have an alerting system which alerts whenever available memory in VM is less than 20%.

So as a quick fix we are clearing cache after each time rsync runs using command echo 2 > /proc/sys/vm/drop_caches.

I'm completely against this, because clearing cache will affect the performance. But on the internet there are few articles which suggests clearing cache after running rsync jobs some of them are 1, 2. Also there are lot of other resources on internet which says not to clear cache.

Considering only rsync

1. Do we really need to worry about aggressive caching it is doing because of which our buff/cache is very high
2. How rest of the world is dealing with this? are you guys clearing cache each time you run rsync

rsync command we are using rsync -aA --delete /... /...

Answer 1

Unused memory is wasted, so any spare memory is considered by the kernel to be available as buffer cache. It can be discarded as soon as necessary should an application need the memory.

If your entire folder is cached then that should mean that comparison between the source files and the destination files is very efficient.

The rsync command itself looks fine. You're not using -H (which can be very memory hungry), and you are usefully retaining timestamps (-t implied by -a). The only other part to be aware of is that as far as rsync is concerned you are copying from one part of the local filesystem to another, so files that just need to be updated will still be copied in their entirety.