rootfs corruptions

[sebt3]

Hello,

I have 5 nano PC T4 boards running at home. Last week 2 boards gave me rootfs corruption. This week I am having the same issue on 2 other nodes...

The last lines written in the syslog before the rootFS went readOnly was :

Jan 3 12:04:59 nano16 kernel: [164560.312121] EXT4-fs error (device mmcblk1p7) in __ext4_new_inode:858: Out of memory
Jan 3 12:04:59 nano16 kernel: [164560.402709] SLUB: Unable to allocate memory on node -1 (gfp=0x2008000)
Jan 3 12:04:59 nano16 kernel: [164560.403325] cache: kmalloc-1024(5653:bf7bcca6b0e64ea0005b131e5094f88131842739e26a01a4c3792ad826c9571e), object size: 1024, buffer size: 1024, default order: 2, min order: 0
Jan 3 12:04:59 nano16 kernel: [164560.404698] node 0: slabs: 0, objs: 0, free: 0

Beside :

root@nano16:/var/log# e2fsck -p /dev/mmcblk1p7
rootfs: recovering journal
e2fsck: Bad magic number in super-block while trying to re-open rootfs

rootfs: ********** WARNING: Filesystem still has errors **********

All the 4 nodes displayed the same symptoms... Now I'm going to fix these FS manually :'( But i'm looking for a fix, anyone ?

[sebt3]

So I found what was the process that was memory balloning causing the low memory issue : flanneld...

Yet this is a kernel bug that should be fixed one way or an other... anyone ?