Re: Serious perfomance issues on fileio. fileio via ramdisk works fine.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hi Renesanso,

On Fri, 2013-03-22 at 12:13 +0400, Фонкин Алексей wrote:
> Hi.
> 
> I use targetcli and LIO Linux stack with FC-to-FC PtP connection
> (target and initiator is QLogic qle2440 ( 02:00.0 Fibre Channel:
> QLogic Corp. ISP2432-based 4Gb Fibre Channel to PCI Express HBA (rev
> 02)  )). 
> UI of targetcli is very simple, all work ok with 3.9 kernel, but I
> have performance troubles. Without buffered=true (fileio) or with it
> sequential r/o is very slow (near 5Mb/s), on random w/o (many files
> copy, not true random) i have 0.04 Mb/s on Windows 7. I can't install
> windows (4% and very fast bilking HDD led on target. From HDD sound I
> can think, that there are aggressive random i/o) in 10 hours period.
> On fedora (run on the same (physical same, not other clone) I have
> sequential  performance near 35 Mb/s, random (many files copy, not
> true random) - 9 Mb/s. Fedora installs and boot (with firmware insert
> in dracut) fine, works ok. 
> 
> If I use file on ramdisk, but not file on HDD in fileio, I have
> performance near 360Mb/s, that very good for me. On target (localhost
> root access) I have 125Mb/s on sequential r/o to file on HDD. Enabling
> buffered=true (fileio) give me 1-2Mb/s more on seq r/o, and near 10%
> on random, but I don't see any cache, buffer, used change (as when I
> use file locally) changes.

What does /sys/kernel/config/target/core/$FILEIO_HBA/$FILEIO_DEV/info
output look like..?  You'll want to check that 'Mode: Buffered-WCE' is
getting enabled after target-restart (see below)

> What I need to do to improve fileio performance on file on HDD? I
> thinks on linux-linux variant there are bottle neck in some detail,
> but in windows variant (I think, but can't prove) there are no or too
> small buffers to vitual FC HDD (via HBA).
> 

The FILEIO default Mode O_DSYNC is probably getting reset if you don't
have the following lio-utils patch in place to save fd_buffered_io=:

http://www.risingtidesystems.com/git/?p=lio-utils.git;a=commitdiff;h=5d0f4829aa130619e81edad3fe0aaa697fa00be4

that adds the ',fd_buffered_io=[0,1]' string to --establishdev calls
based upon active /sys/kernel/config/core/$FILEIO_HBA/$FILEIO_DEV/info,
and should generate the following for /etc/target/tcm_start.sh:

tcm_node --establishdev $FILEIO_HBA/$FILEIO_DEV fd_dev_name=$FILEIO_PATH,fd_dev_size=$FILEIO_SIZE,fd_buffered_io=1

Thanks,

--nab

--
To unsubscribe from this list: send the line "unsubscribe target-devel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html




[Index of Archives]     [Linux SCSI]     [Kernel Newbies]     [Linux SCSI Target Infrastructure]     [Share Photos]     [IDE]     [Security]     [Git]     [Netfilter]     [Bugtraq]     [Yosemite News]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux ATA RAID]     [Linux IIO]     [Device Mapper]

  Powered by Linux