zfs discuss - Feb 2010 - Drive failure causes system to be unusable

Hi!

I have a OSOL box as a home file server. It has 4 1TB USB Drives and 1 TB
FW-Drive attached. The USB devices are combined to a RaidZ-Pool and the FW Drive
acts as a hot spare.

This night, one USB drive faulted and the following happened:

1. The zpool was not accessible anymore
2. changing to a directory on the pool causes the tty to get stuck
3. no reboot was possible
4. the system had to be rebooted ungracefully by pushing the power button

After reboot:

1. The zpool ran in a degraded state
2. the spare device did NOT automatically go online
3. the system did not boot to the usual run level, and no auto-boot zones where
started, GDM did not start either


        NAME         STATE     READ WRITE CKSUM
        tank         DEGRADED     0     0     0
          raidz1-0   DEGRADED     0     0     0
            c21t0d0  ONLINE       0     0     0
            c22t0d0  ONLINE       0     0     0
            c20t0d0  FAULTED      0     0     0  corrupted data
            c23t0d0  ONLINE       0     0     0
        cache
          c18t0d0    ONLINE       0     0     0
        spares
          c16t0d0    AVAIL  



My questions:

1. Why does the system get stuck, when a device faults?
2. Why does the hot spare not go online? (The manual says, that going online
automatically is the default behavior)
3. Why does the system not boot to the usual run level, when a zpool is in a
degraded state at boot time?


Regards,
Martin


-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/zfs-discuss/attachments/20100208/7d7b97ae/attachment.html>

On Feb 8, 2010, at 9:05 AM, Martin Mundschenk wrote:
> Hi!
> 
> I have a OSOL box as a home file server. It has 4 1TB USB Drives and 1 TB
FW-Drive attached. The USB devices are combined to a RaidZ-Pool and the FW Drive
acts as a hot spare.
> 
> This night, one USB drive faulted and the following happened:
> 
> 1. The zpool was not accessible anymore
> 2. changing to a directory on the pool causes the tty to get stuck
> 3. no reboot was possible
> 4. the system had to be rebooted ungracefully by pushing the power button
> 
> After reboot:
> 
> 1. The zpool ran in a degraded state
> 2. the spare device did NOT automatically go online
> 3. the system did not boot to the usual run level, and no auto-boot zones
where started, GDM did not start either
> 
> 
>         NAME         STATE     READ WRITE CKSUM
>         tank         DEGRADED     0     0     0
>           raidz1-0   DEGRADED     0     0     0
>             c21t0d0  ONLINE       0     0     0
>             c22t0d0  ONLINE       0     0     0
>             c20t0d0  FAULTED      0     0     0  corrupted data
>             c23t0d0  ONLINE       0     0     0
>         cache
>           c18t0d0    ONLINE       0     0     0
>         spares
>           c16t0d0    AVAIL  
> 
> 
> 
> My questions:
> 
> 1. Why does the system get stuck, when a device faults?
Are you sure there is not another fault here?  What does "svcs -xv"
show?
 -- richard
> 2. Why does the hot spare not go online? (The manual says, that going
online automatically is the default behavior)
> 3. Why does the system not boot to the usual run level, when a zpool is in
a degraded state at boot time?
> 
> 
> Regards,
> Martin
> 
> 
> _______________________________________________
> zfs-discuss mailing list
> zfs-discuss at opensolaris.org
> http://mail.opensolaris.org/mailman/listinfo/zfs-discuss

Am 08.02.2010 um 20:03 schrieb Richard Elling:
> Are you sure there is not another fault here?  What does "svcs
-xv" show?
Well, I don''t have the result of svcs -xv, since the fault is recovered
by now, but it turned out not to be a hardware failure but an unstable
USB-conectivity. But sill: Why does the system get stuck? Even when a USB-Plug
is unhooked, why does the spare does not go online?

Martin