Aggressive memory leak "kmalloc-128"; BTRFS quotas

[RlndVt]

  

I'm having an aggressive memory leak issue I can't place: Within 5 minutes the 4GB are full. htop and top don't show a massive program growing unbouded. `df` doesn't show a tmpfs that is too large. Only `slabtop` is imforative:

 

  OBJS ACTIVE  USE OBJ SIZE  SLABS OBJ/SLAB CACHE SIZE NAME                   
26638624 26638624  99%    0.12K 832457       32   3329828K kmalloc-128

 

So `kmalloc-128` is growing without bounds. As I understand it kmalloc is a collection of smaller memory allocations? I've tried using a previous kernel version (5.10.16 and 5.10.12) but nada.

 

I've disabled docker, and that along with omv were the only services running. Any suggestions would be great!

[gprovost]

Can you share more info because cannot say much based on the few input you provided

 

ps -o pid,user,%mem,command ax | sort -b -k3 -r


top -b -n 1

 

and the link generated by armbianmonitor -u

 

https://linuxhint.com/check_memory_usage_process_linux/

[RlndVt]

Sadly I seem to have broken my install because the RAM got filled while I was using `armbian-config` to revert a kernel version. With regards to top/ps; As i mentioned nothing pops up with a large usage.

 

I've got a fresh install booting now, I can see if the problem reproduces. (And at what step.)

[RlndVt]

Okay, it has something to do with 'btrfs-cleaner'. After mounting my RAID kmalloc-128 starts growing. Steps I've taken were install, define root password & user name+password, `apt-get update && apt-get upgrade`, reboot and mount my RAID. `umount` also hangs. (Probably because btrfs-cleaner is running.)

 

 

# ps -o pid,user,%mem,command ax | sort -b -k3 -r
  PID USER     %MEM COMMAND
 1381 root      0.4 /usr/bin/python3 /usr/share/unattended-upgrades/unattended-upgrade-shutdown --wait-for-signal
  777 root      0.4 /usr/bin/python3 /usr/bin/networkd-dispatcher --run-startup-triggers
  784 root      0.3 /usr/sbin/NetworkManager --no-daemon
[...]

 

 

# top -b -n 1
top - 13:28:22 up 5 min,  3 users,  load average: 3.83, 2.02, 0.84
Tasks: 179 total,   3 running, 176 sleeping,   0 stopped,   0 zombie
%Cpu(s):  0.7 us, 20.9 sy,  0.0 ni, 78.4 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
MiB Mem :   3799.0 total,   1598.0 free,   1755.0 used,    446.0 buff/cache
MiB Swap:   1899.5 total,   1899.5 free,      0.0 used.   1884.1 avail Mem 

  PID USER      PR  NI    VIRT    RES    SHR S  %CPU  %MEM     TIME+ COMMAND
 2540 root      20   0       0      0      0 R  66.7   0.0   2:32.56 btrfs-cleaner
 2705 root      20   0    8948   3028   2664 R   9.5   0.1   0:00.09 top
 2688 root      20   0    7188   3360   2604 S   4.8   0.1   0:01.15 htop
    1 root      20   0  165100   9372   7052 S   0.0   0.2   0:04.59 systemd
    2 root      20   0       0      0      0 S   0.0   0.0   0:00.03 kthreadd
[...]

 

`armbianmonitor -u`: http://ix.io/2REF

[RlndVt]

After (instantly) disabling quotas the issue is remedied. Even though I have 214 subvolumes it seems ridiculous to me that quotas would use enough RAM to cripple a system. Is this a (upstream) bug?

 

When i read "Performance issues" I think slow IO speed, not crippling RAM usage.