Date: Mon, 7 Mar 2011 19:41:10 +0100 From: Matthias Gamsjager <mgamsjager@gmail.com> To: Joshua Boyd <boydjd@jbip.net> Cc: freebsd-fs@freebsd.org Subject: Re: kmem_map too small with ZFS and 8.2-RELEASE Message-ID: <AANLkTin_aFNercpcBsg41OO3BXL_mYUNn%2BhjncXF14s2@mail.gmail.com> In-Reply-To: <AANLkTin0eZ1_0n1VNYChNTOG6HDayegjHjGeGHso4PMY@mail.gmail.com> References: <1299232133.18671.3.camel@pc286.embl.fr> <20110304100517.GA23249@icarus.home.lan> <AANLkTikQiTi25TR6uDD2umRZQrOL8YZzEC960oWf4wax@mail.gmail.com> <20110304105608.GA23887@icarus.home.lan> <AANLkTimrdnxsxUpmZnr3=w5J8_46ZM91crEfY6c_ZR4z@mail.gmail.com> <20110306090455.GA87055@icarus.home.lan> <AANLkTin0eZ1_0n1VNYChNTOG6HDayegjHjGeGHso4PMY@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Let me too backup my claim with data:
AMD Dual core 4G ram 4x 1TB Samsung drives OS installed on separate ufs disk
FreeBSD fb 8.2-STABLE FreeBSD 8.2-STABLE #0 r219265: Fri Mar 4 16:47:35 CET
2011
loader.conf:
vm.kmem_size="6G"
vfs.zfs.txg.timeout="5"
vfs.zfs.vdev.min_pending=1 #default = 4
vfs.zfs.vdev.max_pending=4 #default= 35
sysctl.conf:
vfs.zfs.txg.write_limit_override=805306368
kern.sched.preempt_thresh=220
Zpool:
NAME STATE READ WRITE CKSUM
storage ONLINE 0 0 0
mirror ONLINE 0 0 0
ad6 ONLINE 0 0 0
ad10 ONLINE 0 0 0
mirror ONLINE 0 0 0
ad4 ONLINE 0 0 0
ad8 ONLINE 0 0 0
NAME SIZE USED AVAIL CAP HEALTH ALTROOT
storage 1.81T 1.57T 245G 86% ONLINE -
Prefetch disable = 1
Version 1.96 ------Sequential Output------ --Sequential Input-
--Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--
--Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec
%CP
fb 10000M 54 74 99180 42 35955 14 140 73 68174 11 180.6
4
Latency 295ms 1581ms 1064ms 428ms 58640us
755ms
Version 1.96 ------Sequential Create------ --------Random
Create--------
fb -Create-- --Read--- -Delete-- -Create-- --Read---
-Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec
%CP
16 6697 39 +++++ +++ 11798 74 10060 61 +++++ +++ 11104
72
Latency 213ms 134us 257us 32866us 2672us
174us
Prefetch disable = 0
Version 1.96 ------Sequential Output------ --Sequential Input-
--Random-
Concurrency 1 -Per Chr- --Block-- -Rewrite- -Per Chr- --Block--
--Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP /sec
%CP
fb 10000M 52 74 107602 46 65443 29 135 74 243760 42
186.5 4
Latency 214ms 865ms 1525ms 79771us 254ms
924ms
Version 1.96 ------Sequential Create------ --------Random
Create--------
fb -Create-- --Read--- -Delete-- -Create-- --Read---
-Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %CP /sec
%CP
16 8152 56 +++++ +++ 4534 36 10966 69 32607 74 9692
71
Latency 112ms 21108us 169ms 30018us 4097us
318us
Read performance 68MB/s vs 243MB/s.
Maybe the kind of workload you have does not work well with prefetch, I
don't know, but for sequential load like I use my NAS for, which is used as
a media tank, it does boost performance quiet a bit.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?AANLkTin_aFNercpcBsg41OO3BXL_mYUNn%2BhjncXF14s2>