Jusa Saari wrote:

Take a look at the "Histogram of requested keys" and compare it to the
"Histogram of successful externally requested keys" in the Node Status
Interface. Notice anything odd ?

I certainly have. The longest bars in successfull are the shortest in the
requests. In other words, my node is much better at getting some keys than
others (which is probably a good thing since it implies specialization)
but these are the keys I get the least requests for (which is a bad thing
and implies serious bugs somewhere). In short, the better I become to
fetch a given key, the less likely it gets routed to me - or so it seems
to me, at least. Can anyone give any more rational explanation to this ?

The attached statistics are very small because I had to restart my node
recently after a period of downtime, and the traffick hasn't really picked
up yet. However, take note of keys d and e - there is less requests for d
than e, despite d having more successes (not just a higher percentage, but
more total successes). The same thing persists, even when there has been
tens of thousands requests.

<snip>
>
Histogram of requested keys.
This count has nothing to do with keys in your datastore
Nov 8, 2003 6:15:56 PM
keys: 73
scale factor: 1.0 (This is used to keep lines < 64 characters)

   0 |===
   1 |=======
   2 |==
   3 |========
   4 |==
   5 |=====
   6 |==
   7 |=====
   8 |=
   9 |====
   a |======
   b |====
   c |===
   d |=====
   e |=========
   f |=======

<snip>
Histogram of successful externally requested keys.
This count has nothing to do with keys in your datastore
Nov 8, 2003 6:15:52 PM
keys: 5
scale factor: 1.0 (This is used to keep lines < 64 characters)

   0 |=
   1 |
   2 |
   3 |
   4 |
   5 |=
   6 |
   7 |
   8 |
   9 |
   a |
   b |
   c |
   d |==
   e |=
   f |


5 datapoints isn't enough data to say anything. You're actually trying to compare the 2 successes on d to the 1 success on e! Get a lot more data (like *at least* 50x more) before you post your evidence. BTW, my node is not showing any obvious sign of the odd behavior you describe.


-Martin


_______________________________________________ Devl mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/devl

Reply via email to