I have a MongoDB 4.4 server (single node, no replicas) for storing IoT-style data. Data is written to several collections every few seconds by my NodeJS app. Documents are not updated/modified, and reads are less common than writes.
I have TTL indices on my big collections so that data older than 6 months is deleted. However, Mongo seems to consume more and more disk space. When the disk inevitably fills up, Mongo and my app stop working. I need to stop Mongo from consuming increasing amounts of disk space.
If I call stats() on my big collections, I can see that there are gigabytes of "file bytes available for reuse". But when I use db.runCommand({compact:'big_collection'}), it doesn't seem to release any space. Other people seem to have similar experiences. I wish I understood why compact isn't working.
I suspect the best alternative approach is to remove the TTL index, and then Cap the collection to a fixed size, but I'd like to hear if anyone has experience with such a process, or alternative recommendations.
