Date: Sun, 22 Jan 2017 17:45:15 +0100 From: Tobias Oberstein <tobias.oberstein@gmail.com> To: freebsd-fs@freebsd.org Subject: Re: stuck on low scrub performance Message-ID: <8b408926-768b-8df9-e38c-b617cce30b16@gmail.com> In-Reply-To: <b0dc09d8-3579-f7c0-f72a-3e2ce798f443@gmail.com> References: <b0dc09d8-3579-f7c0-f72a-3e2ce798f443@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
FWIW, the issue is gone .. it's scrubbing with 130MB/s now:
[oberstet@bvr-file1 ~/scm/parcit/RA/adr/services]$ zpool status tank
pool: tank
state: ONLINE
scan: scrub in progress since Tue Jan 17 11:13:29 2017
55.1T scanned out of 89.4T at 127M/s, 78h42m to go
96K repaired, 61.64% done
Note: the pool consists of 4 disk sets (striped) with 6 disks each
(RAID-Z2).
Am 19.01.2017 um 08:56 schrieb Tobias Oberstein:
> Hi,
>
> I've got a 146TB RAID-Z2 magnetic platter pool which I am scrubbing, but
> performance is very poor:
>
> pool: tank
> state: ONLINE
> scan: scrub in progress since Tue Jan 17 11:13:29 2017
> 5.02T scanned out of 89.4T at 32.2M/s, (scan is slow, no
> estimated time)
> 0 repaired, 5.61% done
>
> I've already tried modifying various knobs, with no success (I changed
> oids while the scrub is running, without restarting - does that take
> effect?).
>
> I've tried to collect as much info as possible using this script
>
> https://gist.github.com/oberstet/c66baf777ea77ae843b64271df7420a3
>
> with this output
>
> https://gist.github.com/oberstet/a8e12227486ebeb1e85f6f56b880ce7d
>
> ---
>
> My expectation from the general performance of this array would be
> scrubbing with 500MB/s (10x more) - this is what we regularily see on
> this array when it goes to the magnetic platters on sequential read loads.
>
> Any helps would be greatly appreciated, as at that speed, the scrubbing
> takes weeks =(
>
> Cheers,
> /Tobias
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?8b408926-768b-8df9-e38c-b617cce30b16>