[OpenIndiana-discuss] Very slow write performance on a 3 disk raidz1-0 zpool

Julius Roberts hooliowobbits at gmail.com
Fri Aug 24 10:24:14 UTC 2012


On 24 August 2012 14:58, Jan Owoc <jsowoc at gmail.com> wrote:
> How are you generating these writes? Are they maybe small files over a
> wireless network? If so, what if you generated the I/O using:
> dd if=/dev/zero of=/Backups/zero.file

seems the slow write performance appears regardless of the workload,
but in the instance there it was the rsync of a 30gb SQL database dump
over gigabit ethernet

> Is this behaviour new, or did you change anything recently? (Did you
> really copy over all those GBs at 2MB/s?) I'm going to rule out
> underpowered system, since even an Atom should be able to do sha256
> checksums and parity calculations faster than 2 MB/s.

this is basically a new instance of OIa5, the performance has always
been slow.  it's an athlon x2 3000+ machine, reasonably powered, and
yes, copies have been taking a LOONG time :( it does only have 1GB of
ram, and i know that will limit rsync performance but really, not down
to 2MB/s surely.  also top shows the following, it's not paging out.

load averages:  0.30,  0.30,  0.29;               up 1+03:46:45

20:21:59
53 processes: 52 sleeping, 1 on cpu
CPU states: 77.2% idle,  1.0% user, 21.8% kernel,  0.0% iowait,  0.0% swap
Kernel: 2625 ctxsw, 32 trap, 2106 intr, 2997 syscall, 4 flt
Memory: 991M phys mem, 86M free mem, 495M total swap, 489M free swap
-- 
Kind regards, Jules

<golgy> whats so wrong with plumb?
<hoolio> nothing, in itself.  it's just for me, knowing what it means
infers i cannot any longer pretend to not be a complete square when it
comes to computers
<Gryphon> I don't know that knowing anything about plumb turns you
into a nerd, but this conversation already has
<hoolio> are you calling me nerdy?
<checkers> hoolio: you know what initramfs means, AND does. You're
lost to the non-geek world already
<Gryphon> yes
<hoolio> hrm
<hoolio> goodbye cruel world.



More information about the OpenIndiana-discuss mailing list