IO Tokens
From LinodeWiki
NOTE: This only applies to UML nodes, not Xen
Each Linode shares I/O bandwidth by trackings I/O Requests. Each time you use I/O (like hitting the disk, including for swapping), you use up io_tokens. When you have no more, your disk access is throttled to the token_refill rate. Tokens are replenished on a regular basis up to a maximum of 2000000.
[edit] Example of io_token output
$ cat /proc/io_status io_count=12475222 io_rate=107 io_tokens=1823 token_refill=512 token_max=2000000
An io_tokens level of 1,823 is very low and you should restrain your use of swap space.
Note... If you can't seem to access your linode (which can sometimes happen if you are seriously low on tokens and sshd can't find enough IO to process the incoming request), you can request the above information via ssh:
ssh yourUser@hostXX.linode.com io_status
replace yourUser with your own user name, and choose the host your linode is on for hostXX.
[edit] The breakdown of the counters
io_count = total amount of tokens used since boot
io_rate = tokens currently being used
io_tokens = number of tokens available to be used
token_refill = number of tokens that io_tokens will increase by per second.
[edit] Token-Limiter README and patches
http://www.theshore.net/~caker/uml/patches/token-limiter.README
