From: thornber@redhat.com
To: "Darrick J. Wong" <darrick.wong@oracle.com>,
Mike Snitzer <snitzer@redhat.com>,
device-mapper development <dm-devel@redhat.com>,
Joe Thornber <ejt@redhat.com>
Subject: Re: Another cache target
Date: Fri, 14 Dec 2012 12:11:44 +0000 [thread overview]
Message-ID: <20121214121143.GD3022@raspberrypi> (raw)
In-Reply-To: <20121214102443.GB3022@raspberrypi>
On Fri, Dec 14, 2012 at 10:24:43AM +0000, thornber@redhat.com wrote:
> I'll add some tests to my test suite that use your maxiops program and
> see if I can work out what's going on.
I've played with your maxiops program, and added these tests to the
suite:
def maxiops(dev, nr_seeks = 10000)
ProcessControl.run("maxiops -s #{nr_seeks} #{dev} -wb 4096")
end
def discard_dev(dev)
dev.discard(0, dev_size(dev))
end
def test_maxiops_cache_no_discard
with_standard_cache(:format => true,
:data_size => gig(1)) do |cache|
maxiops(cache, 10000)
end
end
def test_maxiops_cache_with_discard
size = 512
with_standard_cache(:format => true,
:data_size => gig(1),
:cache_size => meg(size)) do |cache|
discard_dev(cache)
report_time("maxiops with cache size #{size}m", STDERR) do
maxiops(cache, 10000)
end
end
end
def test_maxiops_linear
with_standard_linear(:data_size => gig(1)) do |linear|
maxiops(linear, 10000)
end
end
The maxiops program appears to be doing random writes over the device
(at least the way I'm calling it). So I'm not surprised the mq policy
can't be bothered to cache anything.
Even an agressive write policy wouldn't do much good here, as maxiops
is continuously writing. Such a strategy needs bursty io, so the
cache has time to clean itself.
Discarding the device before running maxiops, as discussed, does
indeed persuade mq to cache blocks as soon as they're hit (see
test_maxiops_cache_with_discard).
As a sanity check I set up the cache device with various amounts of
SSD allocated and timed a short run of maxiops. For a small amount of
SSD, performance is similar to that of my spindle, for as much SSD as
spindle, performance is the same as my SSD.
SSD size | Elapsed time (seconds)
128m | 32
256m | 23
512m | 13.5
1024m | 3.4
Now the bad news is I'm regularly seeing runs that have terrible
performance; not a hang since the io stall oops isn't triggering. So
there's obviously a race in there somewhere that's getting things into
a bad state. Will investigate more, it could easily be an issue in the
test suite.
- Joe
next prev parent reply other threads:[~2012-12-14 12:11 UTC|newest]
Thread overview: 60+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-12-13 20:19 Another cache target Joe Thornber
2012-12-13 20:19 ` [PATCH 1/8] [persistent-data] Fix a bug in btree_del, and another bug that was compensating for it Joe Thornber
2012-12-13 20:19 ` [PATCH 2/8] [persistent-data] dm_btree_walk Joe Thornber
2012-12-13 20:19 ` [PATCH 3/8] [persistent-data] tweak an error message Joe Thornber
2012-12-13 20:19 ` [PATCH 4/8] [dm-bio-prison] Change the bio-prison interface so the memory for the cells is passed in Joe Thornber
2013-01-14 10:02 ` Alasdair G Kergon
2013-01-14 14:06 ` thornber
2013-01-14 14:22 ` Alasdair G Kergon
2013-01-21 23:32 ` Alasdair G Kergon
2013-01-22 11:31 ` thornber
2013-01-22 12:10 ` Alasdair G Kergon
2012-12-13 20:19 ` [PATCH 5/8] [dm-thin] Fix a race condition between discard bios and ordinary bios Joe Thornber
2012-12-14 15:52 ` Mike Snitzer
2013-01-22 0:03 ` Alasdair G Kergon
2013-01-24 2:35 ` Alasdair G Kergon
2013-01-24 13:23 ` thornber
2013-02-06 0:11 ` Mikulas Patocka
2013-02-07 11:20 ` thornber
2012-12-13 20:19 ` [PATCH 6/8] [persistent-data] Add a transactional array Joe Thornber
2013-01-22 21:18 ` Alasdair G Kergon
2013-01-23 12:07 ` thornber
2013-01-25 20:11 ` Alasdair G Kergon
2013-01-28 13:06 ` thornber
2013-01-28 20:25 ` Alasdair G Kergon
2013-01-28 14:57 ` thornber
2013-01-28 20:22 ` Alasdair G Kergon
2012-12-13 20:19 ` [PATCH 7/8] [persistent-data] transactional bitset Joe Thornber
2013-01-22 21:59 ` Alasdair G Kergon
2012-12-13 20:19 ` [PATCH 8/8] [dm-cache] cache target Joe Thornber
2012-12-14 0:17 ` Darrick J. Wong
2012-12-14 10:09 ` thornber
2013-02-12 15:27 ` Alasdair G Kergon
2013-02-12 16:40 ` Alasdair G Kergon
2013-02-12 17:29 ` Alasdair G Kergon
2013-02-14 13:57 ` Joe Thornber
2013-02-14 14:05 ` Joe Thornber
2013-02-14 21:06 ` Alasdair G Kergon
2012-12-13 21:57 ` Another " Mike Snitzer
2012-12-14 1:16 ` Darrick J. Wong
2012-12-14 2:19 ` Mike Snitzer
2012-12-14 2:27 ` Mike Snitzer
2012-12-14 2:42 ` Darrick J. Wong
2012-12-14 4:23 ` Mike Snitzer
2012-12-14 2:34 ` Darrick J. Wong
2012-12-14 10:24 ` thornber
2012-12-14 12:11 ` thornber [this message]
2012-12-14 21:51 ` Darrick J. Wong
2012-12-15 8:23 ` Joe Thornber
2012-12-18 1:49 ` Darrick J. Wong
2012-12-18 2:31 ` Alasdair G Kergon
2013-01-08 0:19 ` Darrick J. Wong
2013-01-08 13:55 ` thornber
2012-12-22 18:50 ` Mark Hills
2012-12-17 16:54 ` Heinz Mauelshagen
2012-12-18 15:44 ` basic cache policy module fix [was: Re: Another cache target] Mike Snitzer
2012-12-20 1:14 ` Darrick J. Wong
2012-12-20 12:57 ` Heinz Mauelshagen
2012-12-20 13:24 ` Mike Snitzer
2012-12-20 16:10 ` Darrick J. Wong
2012-12-20 17:02 ` Heinz Mauelshagen
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20121214121143.GD3022@raspberrypi \
--to=thornber@redhat.com \
--cc=darrick.wong@oracle.com \
--cc=dm-devel@redhat.com \
--cc=ejt@redhat.com \
--cc=snitzer@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.