How about a Linode host statistics page?

Could show the following about each of the hosts:
- Uptime
- Load/CPU Utilization
- IP traffic
- Current running kernel version
- I/O bandwidth (with protocol distribution?)
- Number of Linodes
- Datacenter
- Number of unique IPs
- <anyone others?>

I know that I've wondered about many of these things while troubleshooting problems with my Linode and being able to verify my usage versus the host's usage would have helped many times. I also just love stats and information.

i would like to see something like this made available too.

A lot of that crap can be retrieved from a php script called phpsysinfo

Here as an example: http://www.mikeopolis.biz/sysinfo/

I'm pretty sure that would display information about the system running inside the Linode and not the host the Linodes are running on.

In terms of bogomips, It seems my Linode is better than yours

http://justreadthis.com/sysinfo/

3900 bogomips instead of 3200.
Now, I'm not sure whether this means I'm on a faster host, or simply because I use different kernel. I use 2.6.7-linode3-1um kernel on host31.linode.com 2.6.7-1-bigmem #1 SMP. Whatever it is, 700 bogomips (over 20%) is highly significant

My Bogomips went from 1900 running the 2.4 kernel to 2300 running the 2.6.7 kernel. Don't know why my sysinfo says its localhost.localdomain instead of using the real hostame.

Eh? Only 1900? Is it a Linode? I thought linode hosts share similiar server config? Why is it THAT much different?



That's easy You must have an entry on your /etc/hosts that translates your public IP address (whatever it is) to localhost.localdomain. Otherwise, sysinfo would return whatever your reverse DNS address is.

Yup. Host20. Linode. Its even a Linode 128.



After checking /etc/hosts.. I realized I still had 127.0.0.1 listed as localhost.localdomain as well as another line that had 127.0.0.1 pointed to my hostname. I removed the first line and everything fell into place.

Hmmm ... After checking Linode's offerings I realize it only guarantee CMR of 200+ MHz on your Linode 128. So 2300 bogomips might be normal, if it turns out that my host (host31) has much more Linodes than your host (host20).

But, If you check out http://www.linode.com/products/network.cfm, under server hardware, it SAYS the minimum server hardware is a 2.4 GHz Intel Pentium-4 Processor, 400MHz Front Side Bus. Now THAT's odd. My PC (a 2.4 GHZ P4) has 4718.59 bogomips running kernel 2.6.7. UML running on it will get over 3000 bogomips everytime.

So, your hardware is probably less than a 2.4 GHz P4. What gives? Is Linode promoting false advertising? Or is the advertised hardware for new Linodes only? But again, since the guaranteed CMR is only 200+ MHz you probably can't complain

Any response here, caker?

OTOH, Bogomips are determined when you boot up a system. If the host was doing something, or the load was unusual, when I last booted my host, it would affect the reported bogomips (a Linus invention of course). I should reboot my linode one of the days to see if it gets any higher.

Bogomips do nothing other than report the number of times it took the processor to run a very simple kind of loop. It's a timing calibration thing. It is meaningless when it comes to measuring actual performance of a computer in any way that matters at all (except that there is a general correlation between higher bogomips scores and faster computers, but the granularity is so large that it's not worth worrying about differences of 25% or less).

For the life of me I wish that Linus had never named that measurement "bogomips", there has been so much confusion about this term and false meaning applied to it over the years.

bogomips - the whole term implies it's bogus and full of bogosity. I can't think of a better term

_________________
Rgds
Stephen
(Linux user since kernel version 0.11)

Sigh. No, there is no false advertising done by Linode.com. Please check your facts before making false assumptions. Bogomips are pointless. If at the very moment your Linode is booting, other Linodes are consuming CPU time, then the bogomips rating is going to be affected.

-Chris

I did check my facts, in terms that there is a general correlation between higher bogomips scores and faster computers. I did know that if when UML is booting the CPU is busy duing something else, then the bogomips rating is going to be affected.

I did not know, however, the fact that the values could be THAT much different. I expected only 20-25% difference.

Thank you for your clarification.

BTW, you haven't respond to the initial request yet : various statistics of each host. The load info on Linode's overview page is good, but it will be even better if we can see more.

True, but people seem to focus on the "mips" part of the name, which leads to endless discussion about this useless value.

I think it should have been named something like "timing loop calibration factor". Nobody would compare these values to see whose is "better".

In fact, I think that it's time that Linux dropped reporting of this "bogomips" value altogether. As far as I know it's never been used for anything useful, it's just resulted in confusion.