On Wed, 2007-11-21 at 00:07 -0500, seth vidal wrote:
>  I
> > just took the longest latency there and doubled it to get 5 seconds.
> > 
> >  Anyone have any violent reactions to upping the timeout?
> >  Anyone want a bigger/smaller timeout than 5 seconds?
> 
> your analysis makes sense but I don't know why to go all the way up to 5
> seconds. Given the numbers you're hitting back it seems like 1 or 2
> seconds would be enough.

 Well as I said, I just took the rough average from the two biggest
differences[1] 2.793 and 2.386 and doubled it ... I completely agree it
wasn't scientific :)
 But I also couldn't think of a reason why someone would want to do the
"move to next mirror" operation within five seconds, and it seemed
better to error on the side of stopping everything given the confusion
of the feature itself ... in fact I was almost tempted to suggest 10
seconds, due to [1], but I figured it was possible people might actually
decide something was slow and to move to another mirror within that
time.

 It's true that, in the data I had, a much smaller change seems to do
almost as well statistically:

0.02  0.20  0.27  0.34  0.35
0.36  0.36  0.38  0.38  0.40
0.48  0.54  0.57  0.67  0.72
0.72  0.76  1.00  1.12  2.39
2.79  5.06

...so even 0.75 of a second gives 16 of 22, 73%.


[1] I discounted the 5 second difference because that was mostly me ...
as I usually have to make a mental jump to "oh yeh I need to hit C-c
twice".

-- 
James Antill <[EMAIL PROTECTED]>
Red Hat

Attachment: signature.asc
Description: This is a digitally signed message part

_______________________________________________
Yum-devel mailing list
[email protected]
https://lists.dulug.duke.edu/mailman/listinfo/yum-devel

Reply via email to