[Watford] SLUG

[Yvan Seth]

On Mon, Mar 29, 2010 at 01:53:52PM +0100, Toby Deans wrote:
> Try a DO_REBOOT as root? I'd be surprised if the USB key is failing
> although this can happen - What is hanging on the other USB port? I
> think Magnus had some sort of Hub with a variety of donated discs so
> could they be fried?

The issue is definitely with the storage device that underlies /data -
which is /dev/sda - not some other random device plugged into the
system.

> I'm guessing here and although it's rudimentary, the DO_REBOOT has
> resolved most problems with my boxes.
> 
> I'm away in Mersea over Easter so I won't be at the next LUG (April
> 1st?) - If the box is being too weird we may have to ask Marcus to
> bring it in next meeting. The site and forum seem to still be stable
> for now.

It seems unlikely that a reboot would fix a repeatable sector-specific
storage device issue ... but powercycling computers seems to have
magical properties so is always worth a try. :)

... rebooted ... a pity to destroy our 98 day uptime!

Alas, cat-ing the files from Alain's original email still gives IO
errors.  Worse still, they changed ... which is somewhat weird.

---
cat: czech-windows-1250.inc.php: Input/output error
cat: estonian-utf-8.inc.php: Input/output error
cat: persian-utf-8.inc.php: Input/output error
---

The files in question are unlikely to cause issues to the site unless
someone tries to view it in Persian, Estonian, or Czech (was Slovak
before the reboot) I suppose.

Those specific files aside, from the number of sector errors in dmesg
I'm guessing there must be more files that have this problem.  I'm also
guessing that Alain's backup only recently began to fail on these ...
which means we could have flash that's gradually dying (i.e. if the
flash is something we can replace we should replace it ASAP, we have a
case of phyical bit-rot.)

I don't have time left to try to quatantine and restore files as per
Alain's instructions ... perhaps after work if nobody else gets to it
sooner.

-Yvan