xfs
[Top] [All Lists]

Re: Xfs delaylog hanged up

To: Dave Chinner <david@xxxxxxxxxxxxx>
Subject: Re: Xfs delaylog hanged up
From: Spelic <spelic@xxxxxxxxxxxxx>
Date: Tue, 23 Nov 2010 12:17:41 +0100
Cc: xfs@xxxxxxxxxxx
In-reply-to: <20101122232929.GJ13830@dastard>
References: <4CEAC412.9000406@xxxxxxxxxxxxx> <20101122232929.GJ13830@dastard>
User-agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.11) Gecko/20100713 Thunderbird/3.0.6
On 11/23/2010 12:29 AM, Dave Chinner wrote:
16 disk MD raid-5
What is the storage hardware and the MD raid5 configuration?

Tyan motherboard with 5400 chipset
dual Xeon E5420

16 disks on this one:
05:00.0 RAID bus controller: 3ware Inc 9650SE SATA-II RAID PCIe (rev 01)
        Subsystem: 3ware Inc 9650SE SATA-II RAID PCIe
        Flags: bus master, fast devsel, latency 0, IRQ 16
        Memory at ce000000 (64-bit, prefetchable) [size=32M]
        Memory at d2600000 (64-bit, non-prefetchable) [size=4K]
        I/O ports at 3000 [size=256]
        [virtual] Expansion ROM at d26e0000 [disabled] [size=128K]
        Capabilities: <access denied>
        Kernel driver in use: 3w-9xxx
        Kernel modules: 3w-9xxx

mdadm --create /dev/md5 --raid-devices=16 --chunk=1024 --level=5 --metadata=1.0 pci-0000:05:00.0-scsi-0:0:{15..0}:0 --name=backup:perftest --assume-clean

mount line:
        /dev/mapper/perftestvg-xfslv /perftest/xfs xfs 
rw,relatime,attr2,delaylog,nobarrier,inode64,logbufs=8,logbsize=256k,noquota
0 0
FYI, relatime, attr2, logbufs=8 and noquota are default values that
you don't need to specify.

ok, that was the /proc/mounts line..

mkfs.xfs -l size=128m -d agcount=16 <device>

mount -o delaylog,nobarrier,inode64,logbufs=8,logbsize=256k <device> <mountpoint>


Thanks for the test case - I'll try to reproduce this. One more key
piece of information I need: the output of mkfs.xfs so I can see your
filesystem geometry. if you have the filesystem mounted, then the
output of 'xfs_info<mntpt>' will give the same information.

/# xfs_info /perftest/xfs/
meta-data=/dev/mapper/perftestvg-xfslv isize=256 agcount=16, agsize=50331648 blks
         =                       sectsz=512   attr=2
data     =                       bsize=4096   blocks=805306368, imaxpct=5
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0
log      =internal               bsize=4096   blocks=32768, version=2
         =                       sectsz=512   sunit=0 blks, lazy-count=0
realtime =none                   extsz=4096   blocks=0, rtextents=0

It is stable enough that all my local machines are running with
delaylog. And as a dev team, we have enough confidence in it to
remove the EXPERIMENTAL status for the 2.6.37 release. That doesn't
mean it is perfectly bug free - it just means more people will start
to use it.  And as this happens, we will find new problems in
different environments (just like you have), but this is a pretty
normal occurrence for new code.

ok I understand...

thanks for your work!


<Prev in Thread] Current Thread [Next in Thread>