On Mon, Jun 3, 2019 at 10:36 AM I wrote: > > I just noticed the reported journal size is exactly 1G, which would > account for the difference: > Journal size: 1024M > That's assuming the units are bytes; if they are blocks, it's just a > crazy value. > > I'll see what the extN experts have to say. > Ross
Ted Ts'o over on https://www.spinics.net/lists/linux-ext4/index.html (sorry, reply isn't in index yet) confirms that the big log is the cause of the 3G vs 4G discrepancy, and the big log is a holdover from the original 4TB size I specified. There are some additional suboptimalities from shrinking by a factor of 1000, but the log size is the only substantial one. To reset the journal size he recommended removing and then recreating the journal, which worked for me. df now reports 4G as the volume size, with a corresponding increase in the free data space. root@barley:~#tune2fs -O ^has_journal /dev/vgbarley/cache # remove journal tune2fs 1.44.5 (15-Dec-2018) root@barley:~#date; tune2fs -O has_journal /dev/vgbarley/cache # turn it back on Mon 03 Jun 2019 08:27:37 PM PDT tune2fs 1.44.5 (15-Dec-2018) Creating journal inode: done root@barley:~#date; dumpe2fs -h /dev/vgbarley/cache #.... Inode count: 131072 Block count: 1048576 Reserved block count: 52428 Free blocks: 538855 # was 621488 before, but there's a lot more on it Free inodes: 118435 #..... Journal inode: 8 Default directory hash: half_md4 Directory Hash Seed: 24162063-f4a6-4420-b79b-3ad4f9b71ab7 Journal backup: inode blocks Checksum type: crc32c Checksum: 0x4557d69e Journal features: (none) Journal size: 64M ### <<<<<<<< was 1G Journal length: 16384 Journal sequence: 0x00000001 Journal start: 0 root@barley:~## journal size now 64M instead of 1G! root@barley:~#date; mount /var/local/cache/ Mon 03 Jun 2019 08:28:59 PM PDT root@barley:~#df -h !$ df -h /var/local/cache/ Filesystem Size Used Avail Use% Mounted on /dev/mapper/vgbarley-cache 4.0G 1.9G 1.9G 51% /var/local/cache # only 0.7G in use in my original post; now 1.9G. # the +1.2G is from real files, not overhead Thanks to everyone for your help. Takeaways: * Don't say 4TB when you mean 4GB! (Duh) * e2fs tools report (roughly?) the entire device size for block count, while df only shows the size of the data area, excluding metadata * Mis-sized journals can be fixed by removal and recreation, at least for ext4. Ross
