cgroups.plugin, meaning of "memory usage: file"?


massive packet drops in LXC container after certain period of time
what exactly does Netdata’s cgroups metric “memory usage: file” show?


Debian based LXC with NextPVR in Proxmox VE 7.1, cgroups v2, Netdata v1.29.3

After days of failing to determine the root cause, we finally came across Netdata and it immediately helped us to narrow down the problem area. Thank you very much!!!

The LXC container in question twice used all its assigned memory. We played with the settings in Proxmox, limited the memory to: 2GB, then 6GB, then 1GB (upper chart in screenshot). The packet drops started when available memory was 0, at around 15:25 and 17:00.
The lower chart details it even more: The issue must be around cgroups memory usage “file” which increases dramatically over time (the yellow block).

We checked the plugin documentation, even the man page of cgroups, but couldn’t find the correct section due to our inexperience. Would you mind helping us, what exactly is filling up and how we could identify it even more?

Thanks a lot!

Hi, @mad-sysadmin. That is total_mapped_file value from the memory.stat file. The amount of cached filesystem data mapped with mmap(). Or

of bytes of mapped file (includes tmpfs/shmem)

from Memory Resource Controller — The Linux Kernel documentation