Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 28 Dec 2018 12:18:44 +0200
From:      Andriy Gapon <avg@FreeBSD.org>
To:        Kaya Saman <kayasaman@gmail.com>, freebsd-fs@freebsd.org
Subject:   Re: Upgrade to 12-Release system not booting kernel panic
Message-ID:  <9c5e215a-6cb0-2f97-4652-0bca8683e555@FreeBSD.org>
In-Reply-To: <4a9526c1-2c46-eebe-b92d-d10e522d8941@gmail.com>
References:  <4a9526c1-2c46-eebe-b92d-d10e522d8941@gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On 16/12/2018 09:27, Kaya Saman wrote:
> Hi,
> 
> 
> I've been digging around a little on this one and it seems that there is some
> ZFS wierdness going on.
> 
> 
> My original install was an 11.2-RELEASE fresh install and upon an upgrade to
> 12-RELEASE the system is having issues booting.
> 
> 
> I have a ZFS root pool called zroot; by default this mounts at
> zroot/ROOT/default and a few other non-root / bootable ZFS pools.
> 
> 
> The system is a SuperMicro SC216 chassis with LSI non-RAID HBA.
> 
> 
> I have the boot disks ada0 and ada1 plugged into the rear of the chassis and
> directly into the systemboard which is also a SuperMicro. These drives are both
> Samsung SSD's.
> 
> The 22 drive slots at the front of the chassis are occupied by the other various
> data pools.
> 
> 
> So here is some strangeness... if I remove all 22 drives from the front, the
> system boots fine but straight after boot goes into kernel panic mode and
> reboots before I can even look at the error or get to the login prompt.
> 
> 
> With the non-root pools installed at the BTX loader after scanning through all
> the bios drives I get a bunch of:
> 
> read 264 from ... to 0x...., error 0x10 errors
> 
> then:
> 
> ZFS: i/o error - all block copies unavailable
> 
> ZFS: can't read MOS of pool <non-root-pool-names>

I suspect that this is because of which disks the BIOS presents to the loader
(and in what order).

> after this the system simply hangs?
> 
> 
> I have tried looking around but everything mentioning the MOS error is talking
> about the root pool, a particularly good reference is here:
> http://freebsd.1045724.x6.nabble.com/ZFS-i-o-error-in-recent-12-0-td6245865.html
> 
> 
> In fact I did try to boot with a USB stick and go into Live mode then import all
> the pools on the system. This works without any issue! The pools are fine the
> data is there everything looks normal.
> 
> - I also rebuilt the zpool.cache according to the link just incase there was
> some kind of corruption there, however upon reboot I still get the same issue??
> 
> 
> Looking at a bug report with a kernel panic:
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=220923#c17
> 
> 
> I have attempted to add:
> 
> 
> kern.cam.scsi_delay="50000"
> 
> kern.cam.boot_delay="50000"
> 
> 
> into the /boot/loader.conf file but unfortunately the issue still continues :-(
> 
> 
> I wonder if there is a way to tell to tell the system to only look at certain
> drives for booting??

The boot code looks at all disks presented by the BIOS.
It should look at the disk configured as a boot disk first.

> There is this line in my loader.conf:
> 
> vfs.root.mountfrom="zfs:zroot"
> 
> 
> It maybe the wrong hunch I have but it seems like the system is looking for
> "zroot" on all pools instead of the actual root pool hence the above errors??

You can't look for a pool on all pool, that sounds nonsensical.
The correct way to say this is that the system looks for zroot on all disks.

> Would anyone be able to suggest anything or have any ideas about how to get the
> system back online and booting??

I think that you need to look at the BIOS / disk controller configuration.

-- 
Andriy Gapon



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9c5e215a-6cb0-2f97-4652-0bca8683e555>