On Fri, 25 Jan 2002 kuznet@xxxxxxxxxxxxx wrote:
> > By the way I don't see how you get over wraparound problem.
>
> It needs sampling each sevral seconds to calculate rates,
> so that the problem just does not exist.
This bears repeating: normal-path network
statistics should be used to calculate rates, not absolute values.
Think 'jiffies'. As a human you might interpret a value as "time since
boot" or "number of packets since boot" because you have extra
knowledge, but a program shouldn't make this assumption.
Error counts are a different beast: there you might want to know the
absolute count. But no one expects these to approach 32 bit overflow.
> Well, this attitude has grown as a kind of physiological reaction to
> proposals sort of adding to struct net_device a Cisco-like
> "interface desription" string.
Oooohhh, lets have a /proc/* file that describes the system in text. An
entire book, with only occasional dynamic values.
I do agree with the proposal to add to /proc/net/*
New per-interface statistics files that have only decimal numbers
A program that is only interested in one interface won't trigger
re-reads of all interfaces.
A new (static, read-once) file that describes contains text field labels
The idea to mmap() that statistics file has issues:
We need a timestamp for the reader
We need a way for the reader to trigger a hardware update
(Currently reading /proc/net/dev does this.)
We might need a mechanism so that multiple readers can read
stable/synchronized values.
> Counter wrapping is pseudoproblem of the same nature.
> What will change if netstat will show numbers sort of 761586014605217?
>
> The next step is: "Oh, well, I want to clear them!"
>
> And this will not pass. Clearing kernel counters is hard bug,
> which makes impossible interaction to multiple information consumers.
I used to get a zillion requests (and patches) to clear the counters.
The "multiple readers" answer is the most easily understood response to
this FRF (Frequently Requested mis-Feature).
Donald Becker becker@xxxxxxxxx
Scyld Computing Corporation http://www.scyld.com
410 Severn Ave. Suite 210 Second Generation Beowulf Clusters
Annapolis MD 21403 410-990-9993
|