Date: Mon, 2 May 2011 21:17:18 -0700 From: Jeremy Chadwick <freebsd@jdc.parodius.com> To: Jan Koum <jan@whatsapp.com> Cc: freebsd-fs@freebsd.org, Chris Peiffer <chris@whatsapp.com> Subject: Re: very strange IO issue with FreeBSD 8 and SSD Message-ID: <20110503041718.GA34604@icarus.home.lan> In-Reply-To: <BANLkTinQt4YZiudZUSgxL0x8dJ6MJTueRw@mail.gmail.com> References: <BANLkTin-qEoxxFbjJkDaA_-UZMkza08NNQ@mail.gmail.com> <20110502233601.GA29710@icarus.home.lan> <BANLkTik5tXegwoRvB7XAvpEPb385KjGEtA@mail.gmail.com> <BANLkTinQt4YZiudZUSgxL0x8dJ6MJTueRw@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, May 02, 2011 at 08:53:00PM -0700, Jan Koum wrote:
> On Mon, May 2, 2011 at 8:26 PM, Adam Vande More <amvandemore@gmail.com>wrote:
>
> > On Mon, May 2, 2011 at 6:36 PM, Jeremy Chadwick <freebsd@jdc.parodius.com>wrote:
> >
> >> You might also try comparing iostat output to gstat output, though gstat
> >> refreshes the screen continually making this a little difficult.
> >>
> >
> > gstat -b
> >
>
>
> sure:
>
> $ sudo gstat -b
> dT: 1.007s w: 1.000s
> L(q) ops/s r/s kBps ms/r w/s kBps ms/w %busy Name
> 0 605 33 137 0.4 572 4349 34.7 10.9 ad4
> 0 605 33 137 0.4 572 4349 35.8 10.9 ad4s1
> 0 620 25 149 1.0 595 4280 22.2 9.9 ad5
> 0 605 33 137 0.4 572 4349 36.5 11.0 ad4s1a
> 0 0 0 0 0.0 0 0 0.0 0.0 ad4s1b
> 0 60 18 60 1.1 42 169 2.5 3.6 ad6
> 0 817 30 121 0.2 787 5382 15.9 8.1 ad7
> 0 620 25 149 1.1 595 4280 23.1 10.0 ad5a
> 0 60 18 60 1.1 42 169 2.6 3.7 ad6a
> 0 817 30 121 0.2 787 5382 16.5 8.1 ad7a
To emulate "iostat 1", you will need to run this from inside of a while
loop via the shell. E.g. in sh or bash:
while true; do gstat -b; sleep 1; done
I believe your concern point that started the thread was that
4MBytes/sec was considered bad performance. There are indications from
your iostat output that occasionally the writes are buffered and come in
"in a burst" at 10-11MByte/sec, but your overall average is around
4-5MByte/sec.
You can test your disk I/O by simply dd'ing directly to a file on one of
the filesystems, e.g.
cd /place/where/ad5a/is/mounted
dd if=/dev/zero of=test.bin bs=64k
You can change bs to whatever value you'd like (larger or smaller), but
I tend to stick to 64k (64KBytes). ^C when you're finished, and you'll
see overall I/O statistics. You can run the gstat loop or iostat at the
same time if you wish.
Here's an example:
icarus# dd if=/dev/zero of=test.bin bs=64k
^C4401+0 records in
4400+0 records out
288358400 bytes transferred in 6.575845 secs (43851155 bytes/sec)
Another window running "iostat -x ada0 1":
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 0.0 0.0 0.0 0.0 0 0.0 0
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 0.0 0.0 0.0 0.0 0 0.0 0
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 0.0 61.9 0.0 7924.2 8 19.2 18
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 1.0 334.8 15.9 42790.8 8 19.8 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 1.0 338.5 15.9 43102.6 7 19.7 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 2.0 335.2 31.8 42900.5 8 19.7 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 1.0 336.3 15.9 43047.9 5 20.3 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 1.0 331.7 15.8 42455.8 6 20.3 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 2.0 366.2 31.8 42638.6 8 21.0 100
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 0.0 125.7 0.0 15836.6 0 20.6 37
extended device statistics
device r/s w/s kr/s kw/s wait svc_t %b
ada0 0.0 0.0 0.0 0.0 0 0.0 0
^C
Controller and disk details:
ahci0: <Intel ICH9 AHCI SATA controller> port 0x1c50-0x1c57,0x1c44-0x1c47,0x1c48-0x1c4f,0x1c40-0x1c43,0x18e0-0x18ff mem 0xdc000800-0xdc000fff irq 17 at device 31.2 on pci0
ahci0: [ITHREAD]
ahci0: AHCI v1.20 with 6 3Gbps ports, Port Multiplier supported
ahcich0: <AHCI channel> at channel 0 on ahci0
ahcich0: [ITHREAD]
ada0 at ahcich0 bus 0 scbus0 target 0 lun 0
ada0: <INTEL SSDSA2M040G2GC 2CV102M3> ATA-7 SATA 2.x device
ada0: 300.000MB/s transfers (SATA 2.x, UDMA6, PIO 8192bytes)
ada0: Command Queueing enabled
ada0: 38166MB (78165360 512 byte sectors: 16H 63S/T 16383C)
# camcontrol identify ada0
pass0: <INTEL SSDSA2M040G2GC 2CV102M3> ATA-7 SATA 2.x device
pass0: 300.000MB/s transfers (SATA 2.x, UDMA6, PIO 8192bytes)
protocol ATA/ATAPI-7 SATA 2.x
device model INTEL SSDSA2M040G2GC
firmware revision 2CV102M3
serial number XXX
WWN XXX
cylinders 16383
heads 16
sectors/track 63
sector size logical 512, physical 512, offset 0
LBA supported 78165360 sectors
LBA48 supported 78165360 sectors
PIO supported PIO4
DMA supported WDMA2 UDMA6
media RPM non-rotating
Feature Support Enabled Value Vendor
read ahead yes yes
write cache yes yes
flush cache yes yes
overlap no
Tagged Command Queuing (TCQ) no no
Native Command Queuing (NCQ) yes 32 tags
SMART yes yes
microcode download yes yes
security yes no
power management yes yes
advanced power management no no
automatic acoustic management no no
media status notification no no
power-up in Standby no no
write-read-verify no no
unload yes yes
free-fall no no
data set management (TRIM) yes
I can safely say the conversation is going to immediately turn to "how
does your application work?", including people asking for full source
code and so on. Unless I misunderstand, that's effectively what you're
asking: "why does our application perform so badly on these SSDs?"
--
| Jeremy Chadwick jdc@parodius.com |
| Parodius Networking http://www.parodius.com/ |
| UNIX Systems Administrator Mountain View, CA, USA |
| Making life hard for others since 1977. PGP 4BD6C0CB |
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110503041718.GA34604>