XFS write cache flush policy

[Matthias Schniedermeyer]

On 11.12.2012 11:25, Dave Chinner wrote:
> On Mon, Dec 10, 2012 at 10:45:11PM +0100, Matthias Schniedermeyer wrote:
> > > >> Did you run repair in between the shutdown and remount?
> > > > 
> > > > No.
> > > > 
> > > > XFS (dm-3): Mounting Filesystem
> > > > XFS (dm-3): Starting recovery (logdev: internal)
> > > > XFS (dm-3): Ending recovery (logdev: internal)
> > > > 
> > > >> How many files in that 200GB of data?
> > > > 
> > > > At 0.9GB/file at least 220.
> > > > 
> > > >> http://xfs.org/index.php/XFS_FAQ#Q:_What_information_should_I_include_when_reporting_a_problem.3F
> 
> So I can make any sense of the errors, can you please post the
> rest of the information this link asks for?

- kernel version (uname -a)
For the reported case:
Linux zz 3.6.2 #128 SMP x86_64 GNU/Linux
Userspace is 32bit.

- xfsprogs version (xfs_repair -V)
xfs_repair version 3.1.7

- number of CPUs
1 Socket, 4 cores, HT (Core i7 3770)

- contents of /proc/meminfo
Too late for that
- contents of /proc/mounts
Too late for that 2:
But as all the external mounted HDDs are mounted identical so it would 
be like this:
/dev/mapper/zz /misc/zz xfs rw,noatime,attr2,allocsize=64k,noquota 0 0

- contents of /proc/partitions
To late for that 3:

The relevant entries would be been like this:
   8       80 2930266584 sdf
   8       81 2147482623 sdf1
 254        5 2930266580 dm-5

- RAID layout (hardware and/or software)
- LVM configuration
No RAID or LVM.
A dm-crypt layer though

keyfile=<file with the 65 encryption keys in clear text>
keysize=128
offset=8
device=/dev/sdf
map=zz
cryptsetup loopaesOpen --key-file=$keyfile --key-size 128 --offset $offset $device $map

I don't use a partition but the whole disc except for the first 8 
sectors/4k.

- type of disks you are using
Western Digital WD30EZRX SATA drive.
Connected by SuperSpeed USB3-enclosure.

- write cache status of drives
This is from another one, it's an identical model & enclosure.
write-caching =  1 (on)
Interesting that hdparm -W works over USB.

- size of BBWC and mode it is running in
None

- xfs_info output on the filesystem in question
Technically this is from another drive, but i had formated it with 
identical options:
mkfs.xfs -l size=1024b -s size=4096 ...

- snip -
meta-data=/dev/mapper/zz         isize=256    agcount=32, agsize=22892708 blks
         =                       sectsz=4096  attr=2
data     =                       bsize=4096   blocks=732566645, imaxpct=5
         =                       sunit=0      swidth=0 blks
naming   =version 2              bsize=4096   ascii-ci=0
log      =internal               bsize=4096   blocks=1024, version=2
         =                       sectsz=4096  sunit=1 blks, lazy-count=1
realtime =none                   extsz=4096   blocks=0, rtextents=0
- snip -

- dmesg output showing all error messages and stack traces 
See previous mail.





-- 

Matthias