[Server-devel] anaconda deletes /fsckoptions on F9 based XS

Martin Langhoff martin.langhoff at gmail.com
Wed Sep 17 20:49:22 EDT 2008


On Thu, Sep 18, 2008 at 3:43 AM, Jeremy Katz <katzj at redhat.com> wrote:
> How often are you actually getting to having fsck questions?

Hi Jeremy,

good question. In the field, the XS machines never get switched off -
they are headless, and set in the bios to auto-switch-on. Most of them
will be in locations with unreliable power - so they will switch off
when power gets cut.

Things may evolve so that they have small UPS-style battery packs to
support a controlled shutdown,specially to reduce the wear on the HW.
But for the time being - tough!

When an XS is brought down in the middle of a big write (an this is a
normal thing to happen in our scenario - so we are coding everything
very defensively) it will want to fsck on boot. And there's noone to
hit Y there -- the machine is headless, locked up in a closet. So
without this fsck will literally DoS the machine :-/

What if fsck makes a mess? Well, tough! If the disk is hosed so much
that fsck is going to make a mess, they weren't going to get a usable
XS. The team looking after the XS machines doesn't have the resources
to undertake a big data recovery job. Like most of us, either fsck
knows what it's doing, or we're fsck'd anyway :-/

We'll also provide a backup mechanism for XS - though I'm not sure how
much usage it'll see.

> It's not deleted by anaconda, it's deleted on boot by rc.sysinit.  See
> line 723 or so.

thanks for the tip!




m
-- 
 martin.langhoff at gmail.com
 martin at laptop.org -- School Server Architect
 - ask interesting questions
 - don't get distracted with shiny stuff - working code first
 - http://wiki.laptop.org/go/User:Martinlanghoff


More information about the Server-devel mailing list