Xen users - Feb 2012 - Save options for XFS over ISCSI ~ 17TB (for backup)

hi,

I want to migrate away from Solaris with ZFS to Linux with XFS (BTRFS is to
young) . At the moment, I have Solaris 10 with ISCSI initiator running and a
zpool with a size round about ~17TB. The ISCSI target  is a Linux Debian with
"tgt" Daemon and connected via SAS (LSI Logic / Symbios Logic SAS2008)
to several LSI JBODS (630j).

Every jbod has 12 (2TB) disks and managed by mdadm as a RAID6 device and on top
LVM. I have two ISCSI target servers (one for _cold_ standby) and all servers
has  power connection via UPS and a second direct one.

As I sad, I want to get away from Solaris 10 and know I search for options to
use XFS with save options, in case of emergency (poweroff etc.). We use the
JBODS mostly for backups (bacula) so, performance/realtime isn''t
imported, since network is the limit.

Under ZFS I can make sure,  with zpool scrub (file system check), that all data
are completely  written and consistence. But what is the best way under XFS?
Should I remount XFS every month for a file system check (re/mount) ? What
happens in the case, of ISCSI problems?

So, I need just a few suggestions :-)

cu denny

_______________________________________________
Xen-users mailing list
Xen-users@lists.xen.org
http://lists.xen.org/xen-users

On Fri, Feb 24, 2012 at 2:43 PM, Denny Schierz <linuxmail@4lin.net>
wrote:
> hi,
>
> I want to migrate away from Solaris with ZFS to Linux with XFS (BTRFS is to
young) . At the moment, I have Solaris 10 with ISCSI initiator running and a
zpool with a size round about ~17TB. The ISCSI target  is a Linux Debian with
"tgt" Daemon and connected via SAS (LSI Logic / Symbios Logic SAS2008)
to several LSI JBODS (630j).
... and how does this relate to xen? Are you sending to the wrong list
by any chance?
>
> Every jbod has 12 (2TB) disks and managed by mdadm as a RAID6 device and on
top LVM. I have two ISCSI target servers (one for _cold_ standby) and all
servers has  power connection via UPS and a second direct one.
>
> As I sad, I want to get away from Solaris 10 and know I search for options
to use XFS with save options, in case of emergency (poweroff etc.). We use the
JBODS mostly for backups (bacula) so, performance/realtime isn''t
imported, since network is the limit.
>
> Under ZFS I can make sure,  with zpool scrub (file system check), that all
data are completely  written and consistence. But what is the best way under
XFS? Should I remount XFS every month for a file system check (re/mount) ? What
happens in the case, of ISCSI problems?
>
> So, I need just a few suggestions :-)
So you need the equivalent of "zpool scrub"?
I don''t use xfs, but btrfs has scrub command (at least if you use
latest btrfs-progs from git). Or you can use zfsonlinux. If you''re
interested in either of those two, and need some help, you can ask for
further information in their corresponding list.

-- 
Fajar

On 02/24/2012 08:43 AM, Denny Schierz wrote:
> hi,
>
> I want to migrate away from Solaris with ZFS to Linux with XFS (BTRFS is to
young) . At the moment, I have Solaris 10 with ISCSI initiator running and a
zpool with a size round about ~17TB. The ISCSI target  is a Linux Debian with
"tgt" Daemon and connected via SAS (LSI Logic / Symbios Logic SAS2008)
to several LSI JBODS (630j).
>
> Every jbod has 12 (2TB) disks and managed by mdadm as a RAID6 device and on
top LVM. I have two ISCSI target servers (one for _cold_ standby) and all
servers has  power connection via UPS and a second direct one.
>
> As I sad, I want to get away from Solaris 10 and know I search for options
to use XFS with save options, in case of emergency (poweroff etc.). We use the
JBODS mostly for backups (bacula) so, performance/realtime isn''t
imported, since network is the limit.
>
> Under ZFS I can make sure,  with zpool scrub (file system check), that all
data are completely  written and consistence. But what is the best way under
XFS? Should I remount XFS every month for a file system check (re/mount) ? What
happens in the case, of ISCSI problems?
What''s the problem with zfs? Why don''t you use zfsonlinux?
AFAIK xfs online check cannot be done on xfs.

tamas

Hi,

XFS is of course checked online, too, like almost any journaling filesystem.
Let me add that the BIG HUGE OBVIOUS problem in your setup is the Debian
iSCSI target with the md raid6''s, not the ZFS box.

Florian

2012/2/24 Alexandre Chapellon <a.chapellon@horoa.net>
>  Hello,
>
> Well, this is not truly a question for the Xen mailing list... however.
> If perfromance doesn''t matter that much, you can first use the
"sync" mountoption wich would make sure I/O are written to the media
asap. Also using
> "barriers=yes" (which should be the default) would ensure better
> reliability. Ihave also heard using a different local device as the
> logdevice is a good idea... for exemple a local flash or sata drive.
>
> Regards.
>
> Le 24/02/2012 08:43, Denny Schierz a écrit :
>
> hi,
>
> I want to migrate away from Solaris with ZFS to Linux with XFS (BTRFS is to
young) . At the moment, I have Solaris 10 with ISCSI initiator running and a
zpool with a size round about ~17TB. The ISCSI target  is a Linux Debian with
"tgt" Daemon and connected via SAS (LSI Logic / Symbios Logic SAS2008)
to several LSI JBODS (630j).
>
> Every jbod has 12 (2TB) disks and managed by mdadm as a RAID6 device and on
top LVM. I have two ISCSI target servers (one for _cold_ standby) and all
servers has  power connection via UPS and a second direct one.
>
> As I sad, I want to get away from Solaris 10 and know I search for options
to use XFS with save options, in case of emergency (poweroff etc.). We use the
JBODS mostly for backups (bacula) so, performance/realtime isn''t
imported, since network is the limit.
>
> Under ZFS I can make sure,  with zpool scrub (file system check), that all
data are completely  written and consistence. But what is the best way under
XFS? Should I remount XFS every month for a file system check (re/mount) ? What
happens in the case, of ISCSI problems?
>
> So, I need just a few suggestions :-)
>
> cu denny
>
>
>
> _______________________________________________
> Xen-users mailing listXen-users@lists.xen.orghttp://lists.xen.org/xen-users
>
>
> --
>  <http://www.horoa.net>
>
> Alexandre Chapellon
>
> Ingénierie des systèmes open sources et réseaux.
> Follow me on twitter: @alxgomz <http://www.twitter.com/alxgomz>
>
> _______________________________________________
> Xen-users mailing list
> Xen-users@lists.xen.org
> http://lists.xen.org/xen-users
>


-- 
the purpose of libvirt is to provide an abstraction layer hiding all xen
features added since 2006 until they were finally understood and copied by
the kvm devs.



_______________________________________________
Xen-users mailing list
Xen-users@lists.xen.org
http://lists.xen.org/xen-users

hi,

arrgh, sorry for that. This mail was for XFS-users :-)

Just ignore ... sorry for that one.

cu denny

_______________________________________________
Xen-users mailing list
Xen-users@lists.xen.org
http://lists.xen.org/xen-users