Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 28 Dec 2008 16:14:26 +0200
From:      Danny Braniss <danny@cs.huji.ac.il>
To:        "Erik Scholtz, ArgonSoft GmbH" <e.scholtz@argonsoft.de>
Cc:        freebsd-scsi@freebsd.org
Subject:   Re: Problem with disklabel and filesystem over iSCSI 
Message-ID:  <E1LGwPS-000B6h-Dx@kabab.cs.huji.ac.il>
In-Reply-To: <4957824C.1000601@argonsoft.de> 
References:  <4956B01B.3000509@argonsoft.de> <E1LGrjK-0007Zn-5K@kabab.cs.huji.ac.il> <49575045.4090601@argonsoft.de> <E1LGti5-0009Dv-Uo@kabab.cs.huji.ac.il> <49577118.1080700@argonsoft.de> <E1LGvW7-000ALB-Sl@kabab.cs.huji.ac.il> <4957824C.1000601@argonsoft.de>

next in thread | previous in thread | raw e-mail | index | archive | help
> Danny Braniss wrote:
> >> ok, i started the system in single-user-mode and brought up the iSCS=
I=20
> >> connection. I tried to install the ufs first, after each step I sync=
ed.=20
> >>   No change - problem remains the same (newfs is failing).
> >>
> >> Then I installed the ZFS again, also synced after each step, then=20
> >> unmounted the iSCSI device, synced again (always at least three time=
s).=20
> >> Then rebooted the system (going to single-user-mode again) and the Z=
FS=20
> >> is corrupted again with the same message.
> >>
> > in the case of zfs, you have to
> > 	 /etc/rc.d/zfs stop
> >
>=20
> I used the =22zfs umount tank=22 command. I now tried it with =22/etc/r=
c.d/zfs=20
> stop=22 (both cases: before and after umounting the volume via =22zfs=20
> umount=22) and synced again. Still the effect remains the same: After=20
> reboot the filesystem is corrupted and unreadable.
>=20
> >> I don't think the buffers are the problem.
> >>
> > are you sure that the iscsi target is only used by the freebsd host?
>=20
> 100% for sure. The SAN is (at the moment) connected via Cross-Cable=20
> directly to the FreeBSD-Box, so has only a dedicated connection. Before=
=20
> that it was connected over a seperate Cisco 520G Switch, only used by=20
> the SAN and the FreeBSD Box.
> The networkcard is a INTEL 1GBit. Also tried it with a nVidia nForce Pr=
o=20
> 3400 (also 1GBit). I also played with the MTU sizes: at the moment it i=
s=20
> default (1500) - but other sizes (jumbo size 9000) also do not change=20
> anything (I made sure, both sides (SAN and FreeBSD) uses the same size)=
.

Since you seem to be in a cooperative mood :-), can you try checking
with a smaller volume? say 200GB? either UFS or ZFS?
the biggest I have access to is 921595 Meg and it does not show your prob=
lems.
also, what is the Media Sector size?
fdisk da0 should show it.

cheers,
	danny





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?E1LGwPS-000B6h-Dx>