IRC channel logs

<rekado>looks like we’re running out of inodes on the shared Guix installation at the MDC

<rekado>we use ext4; perhaps that was not a good choice

<dstolfa>rekado: this seems to be a common theme with ext4 + guix

<dstolfa>every month someone will say they've run out of inodes

<rekado>we’re seeing this in the logs:

<rekado>[22969.725651] EXT4-fs warning (device sdc1): ext4_dx_add_entry:2202: Directory (ino: 609) index full, reach max htree level :2

<rekado>[22969.725654] EXT4-fs warning (device sdc1): ext4_dx_add_entry:2206: Large directory feature is not enabled on this filesystem

<rekado>[22969.728546] EXT4-fs warning (device sdc1): ext4_dx_add_entry:2202: Directory (ino: 609) index full, reach max htree level :2

<rekado>[22969.728548] EXT4-fs warning (device sdc1): ext4_dx_add_entry:2206: Large directory feature is not enabled on this filesystem

<rekado>we haven’t run out of inodes yet

<rekado>df -i | egrep 'Filesystem|gnu'

<rekado>Filesystem Inodes IUsed IFree IUse% Mounted on

<rekado>/dev/sdc1 132644864 18304004 114340860 14% /gnu

<rekado>and we’ve got these features: has_journal ext_attr resize_inode dir_index filetype needs_recovery extent 64bit flex_bg large_dir sparse_super large_file huge_file uninit_bg dir_nlink extra_isize

<rekado>apparently, ‘large_file’ was only just activated. I thought this cannot be changed later.

IRC channel logs

2021-09-12.log