Misestimated free RAM about earlyoom

[rfjakob]

Hmmm, could you please post /proc/meminfo ? Old "free" versions misaccount tmpfs!

Edit: Recent "free" has an extra column "available" - equivalent to "MemAvailable" in /proc/meminfo. And it has dropped the "-/+" row because it is inaccurate on modern kernels due to tmpfs.

$ free -m
              total        used        free      shared  buff/cache   available
Mem:           7841        2243        2597        1175        3000        4066
Swap:          1023         499         524

[rfjakob]

Ok I think I can calculate it from the "free" ouput as well: 15GB are "cached". But 13GB are "shared", an this usually means files in tmpfs. These 13GB must be deducted from the " cached" number because the memory is not actually idisposable. From the 19GB "free", this takes us to 6GB available. Earlyoom stated you have 3GB available, the difference is probably from the virtualbox vm that was killed?

[kernc]

$ cat /proc/meminfo
MemTotal:       32359008 kB
MemFree:         2657452 kB
MemAvailable:    6309220 kB
Buffers:          339344 kB
Cached:         16948472 kB
SwapCached:            0 kB
Active:         23319752 kB
Inactive:        3883036 kB
Active(anon):   21424576 kB
Inactive(anon):  2613048 kB
Active(file):    1895176 kB
Inactive(file):  1269988 kB
Unevictable:          32 kB
Mlocked:              32 kB
SwapTotal:             0 kB
SwapFree:              0 kB
Dirty:               392 kB
Writeback:             0 kB
AnonPages:       9915012 kB
Mapped:          2453228 kB
Shmem:          14122652 kB
Slab:             876384 kB
SReclaimable:     712060 kB
SUnreclaim:       164324 kB
KernelStack:       14240 kB
PageTables:        83920 kB
NFS_Unstable:          0 kB
Bounce:                0 kB
WritebackTmp:          0 kB
CommitLimit:    16179504 kB
Committed_AS:   30452072 kB
VmallocTotal:   34359738367 kB
VmallocUsed:           0 kB
VmallocChunk:          0 kB
HardwareCorrupted:     0 kB
AnonHugePages:         0 kB
HugePages_Total:       0
HugePages_Free:        0
HugePages_Rsvd:        0
HugePages_Surp:        0
Hugepagesize:       2048 kB
DirectMap4k:     1738240 kB
DirectMap2M:    30167040 kB
DirectMap1G:     2097152 kB

Indeed, I have a large tmpfs

$ df -h | grep tmpfs
tmpfs           6.2G  162M  6.1G   3% /run
tmpfs            16G  9.8G  5.7G  64% /dev/shm
tmpfs           5.0M  4.0K  5.0M   1% /run/lock
tmpfs            16G     0   16G   0% /sys/fs/cgroup
tmpfs           9.3G  2.6G  6.8G  28% /tmp
tmpfs           3.1G  4.0K  3.1G   1% /run/user/120
tmpfs           3.1G  628K  3.1G   1% /run/user/1000

VirtualBox was assigned some 3GB.

So you are saying I need more RAM?

[rfjakob]

The line

MemAvailable:    6309220 kB

tells us that you have 6.3 GB available. Once this drops below 3.1 GB earlyoom will start killing processes. What you can do:

1. Deleting the files from /tmp will make the memory available.
2. Start earlyoom with -m 5 so it will start killing processes when 5% of memory are free (default is 10%)
3. Add a few GB of swap

[rfjakob]

But yes, this box is close to running out of memory.

[kernc]

Ah, indeed, I had lots of stale data in /dev/shm which my DE's memory monitor widget oddly doesn't seem to count. Thanks!