So I have been testing a small machine with dedup for use with a few VMs on centos7 using kvm/libvirtd/qemu.
system info:
CPU E5-1620 v3 @ 3.50GHz
125GB RAM
zpool = 2x 500GB HDDs
pool: oddly1
state: ONLINE
scan: scrub repaired 0B in 1h32m with 0 errors on Tue Dec 3 05:38:34 2019
config:
NAME STATE READ WRITE CKSUM
oddly1 ONLINE 0 0 0
mirror-0 ONLINE 0 0 0
wwn-0x5000c500501c03e6 ONLINE 0 0 0
wwn-0x50014ee2b313479c ONLINE 0 0 0
errors: No known data errors
dedup: DDT entries 1445627, size 826B on disk, 266B in core
bucket allocated referenced
______ ______________________________ ______________________________
refcnt blocks LSIZE PSIZE DSIZE blocks LSIZE PSIZE DSIZE
------ ------ ----- ----- ----- ------ ----- ----- -----
1 1.37M 175G 175G 175G 1.37M 175G 175G 175G
2 8.56K 1.07G 1.07G 1.07G 18.2K 2.28G 2.28G 2.28G
4 136 17M 17M 17M 556 69.5M 69.5M 69.5M
8 19 2.38M 2.38M 2.38M 228 28.5M 28.5M 28.5M
16 10 1.25M 1.25M 1.25M 235 29.4M 29.4M 29.4M
32 4 512K 512K 512K 168 21M 21M 21M
64 2 256K 256K 256K 136 17M 17M 17M
128 2 256K 256K 256K 382 47.8M 47.8M 47.8M
256K 1 128K 128K 128K 433K 54.2G 54.2G 54.2G
Total 1.38M 176G 176G 176G 1.81M 232G 232G 232G
-
[root@workstationfed01 ~]# zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
oddly1 464G 178G 286G - 56% 38% 1.31x ONLINE -
How do you calculate how much is cached data vs how much is metadata … the arc_summary command?