I have an NFS mount to a FreeNAS system, and it's used to rsync backup a remote server. When the rsync is finished we have a ton (49G) of inact memory: last pid: 67711; load averages: 6.90, 5.89, 5.85 up 3+16:00:34 10:50:16 76 processes: 9 running, 67 sleeping CPU: 0.0% user, 48.0% nice, 0.1% system, 0.0% interrupt, 51.9% idle Mem: 300M Active, 49G Inact, 12G Wired, 1655M Buf, 748M Free ARC: 5793M Total, 2663M MFU, 644M MRU, 1168K Anon, 131M Header, 2354M Other Swap: 96G Total, 194M Used, 96G Free when I umount the share, it all goes to Free: borg.lerctr.org /home/ler $ sudo umount /vault/backup/TBH Password: last pid: 67724; load averages: 6.57, 6.09, 5.93 up 3+16:03:23 10:53:05 76 processes: 9 running, 67 sleeping CPU: % user, % nice, % system, % interrupt, % idle Mem: 300M Active, 173M Inact, 10G Wired, 1655M Buf, 50G Free ARC: 5794M Total, 2663M MFU, 645M MRU, 1424K Anon, 131M Header, 2354M Other Swap: 96G Total, 194M Used, 96G Free 1) is this expected? 2) if not, how can we prevent/fix it? borg.lerctr.org /home/ler $ uname -aKU FreeBSD borg.lerctr.org 11.0-CURRENT FreeBSD 11.0-CURRENT #4 r284983: Tue Jun 30 18:01:26 CDT 2015 root@borg.lerctr.org:/usr/obj/usr/src/sys/VT-LER amd64 1100077 1100077 borg.lerctr.org /home/ler $ /etc/fstab: freenas.lerctr.org:/mnt/data/TBH /vault/backup/TBH nfs rw 0 0
All the inactive memory is the cache of the files on NFS mount.