power failure, boot, and fsck

[Daniel Feenberg]

On Mon, 9 Jul 2012, Matthew Seaman wrote:

> On 09/07/2012 04:22, Patrick Donnelly wrote:
>> UFS: /dev/ad10s3f (/usr)
>> Automatic file system check failed, help!
>> error aborting boo (sending sigtem to parent)!
>> init: /bin/sh on /etc/rc terminated abnormally, going to single user mode.
>> enter full pathname of shell or RETURN for /bin/sh:
>>
>> In single-user mode I just `fsck /dev/da0s1a` and reboot. That fixes
>> the problem. However, I would like this to be automatic on boot. It
>> would be annoying if I'm out-of-town and the server cannot recover
>> without my help. Any tips?
>
> fsck does run automatically when a filesystem does not get shut down
> cleanly.  However, fsck cannot fix all of the problems a filesystem can
> experience without risk of loss of data.  In those cases, there is no
> option but to stop and ask the operator to intervene.

Won't soft updates solve this problem?

   http://www.freebsd.org/doc/en/books/handbook/configtuning-disk.html

The handbook says. "We recommend to use Soft Updates on all of your file 
systems." but doesn't mention booting specifically. This isn't something I 
have tried (we boot over the network).

>
> Your best bet is to avoid an unclean shutdown entirely.  Buy a UPS.
>

We have lots of UPS systems. They constitute a "single point of failure", 
a prodigous amount of hazardous waste every couple of years. I'd sure like 
to drop them - and not on my foot.

I should say that we stopped using soft updates because the background 
fsck was very slow, but that was on very large partitions. On a boot
drive with no user data, the timing would be fine.

dan feenberg

> 	Cheers,
>
> 	Matthew
>
> -- 
> Dr Matthew J Seaman MA, D.Phil.                   7 Priory Courtyard
>                                                  Flat 3
> PGP: http://www.infracaninophile.co.uk/pgpkey     Ramsgate
> JID: matthew at infracaninophile.co.uk               Kent, CT11 9PW
>
>
>
>