For the last 5 hours, any hardrive I/O seems to absolutely ages. Caker, could you please look into it?

Copying a simple 10Mb file seems to take more than 15 minutes!

cat /proc/io_status

and post the output here

_________________
Jay Faulkner
http://oldos.org

io_count=2321319 io_rate=96 io_tokens=-26 token_refill=100 token_max=400000

This was just after I upgraded Debian Woody -> Sarge on the linode.

You have hit the i/o rate limiter.

Your token consumption (io_rate=96) is almost equal to the refill rate (token_refill=100) (and has been higher), while your available tokens (io_tokens=-26) are actually negative, as opposed to a maximum value (quiescent system) of 400000 (token_max).

Maybe your Linode thrashed its swap space during the upgrade - if so, it will take a few hours to recover or you can raise a ticket and ask caker to reset io_tokens.

_________________
/ Peter

I don't think it ever recovers by itself. I hit the limiter going to Sarge a while back, and waited, and was still on the low limit. I created a support ticket and caker helped me out.

Somebody later said that rebooting your Linode resets the limit, so you might try that.

I would look at processes running with "ps aux" There is probably a hung proccess eating I/O.

_________________
-- Surferdude

I'd usually suggest just Rebooting and praying no critical data is lost, but considering your system is in the middle of updating potentially critical files - don't do it!

Don't login via SSH, use LISH instead to save a few io_tokens. Start killing whatever process you can safely kill, and try to bring the beast down gently.

Make every command you run count because I've had simple commands like "ps aux" take minutes on a system that had hit the limiter.

...or just let it chug away on the upgrade for the next couple weeks and probably annoy everyone else on your Linode.

Isn't he done with the upgrade?



I'd just reboot.