(this post is purely off-topic to real discussion. I'm running the
latest unstable build on a k6-3 400 cpu with 384 megs ram.. bandwidth is
3000/400 (upstream maxes at out 50K/sec))

Yesterday some people were toying around with the idea of setting
maxlog2DataSize to 16 in irc.  Heres the conf description..

> # The logarithm to the base 2 of the maximum file data+metadata size
> # that the node will accept. 20 means 1 megabyte, which is reasonable.
> maxLog2DataSize=18

I've changed my to 18 since the majority of my DS files are 256k.  After
13 hours of operation I have a few interesting numbers, maybe someone
would care to explain if it makes any sense... I have some ideas but
they're most likely wrong.

Connections open (Inbound/Outbound/Limit) 512 (318/194/512) 
Connections transferring (Transmitting/Receiving) 470 (396/74) 
Data waiting to be transmitted/received 144 MiB/61 MiB 
Amount of data transmitted/received over currently open connections
45MiB/82 MiB 
Total amount of data transmitted/received 229 MiB/472 MiB 
Outbound connections that are to peers not in the routingtable 50.51546%
Number of distinct nodes connected 102

This is interesting because I've received twice as much as I've
transmitted, which is totally opposite to normal operation if I were
handling 1 meg keys..    If this were after 13 hours of normal operation
I would have likely transmitted 1.5 - 2.0 gigs and received half of
that.  All the numbers here are completely opposite of what I normally
see.  

This node really shouldn't be waiting to send data as it has plenty of
upstream bandwidth to take care of it.

470 connections transferring, yikes ? *shrug* and there's only 102
distinct nodes (normally I would see 275-325 distinct nodes at this
point....) In the OCM it lists some 15+ connections opened to a bunch
individual nodes (which I've never seen before.)

http://soros.ca/ocm.html
 
-----------------------------------------------------------
It also seems to be having problems with open files somewhere, although
I'm not sure where or why its happening, except that my terminal was
full of these messages overnight ( a few hundred or thousand at least..
I've never seen this before either.)

Caught a java.io.IOException: Too many open files,
LSL.processConnections failing
java.io.IOException: Too many open files
        at sun.nio.ch.ServerSocketChannelImpl.accept0(Native Method)
        at
sun.nio.ch.ServerSocketChannelImpl.accept(ServerSocketChannelImpl.java:134)
        at
freenet.transport.ListenSelectorLoop.processConnections(ListenSelectorLoop.java:104)
        at
freenet.transport.AbstractSelectorLoop.loop(AbstractSelectorLoop.java:729)
        at
freenet.transport.ListenSelectorLoop.run(ListenSelectorLoop.java:146)
        at java.lang.Thread.run(Thread.java:536)

------------------------------------------------------------------
The percentages are nothing special, the node seems to be handling
slightly more requests though, 

localQueryTraffic
The amount of queries received, and the number that are not rejected.
Aggregated over every minute . Recorded aggregates over hour:
11/25/03 11:00:00 PM EST 28635 9923 0.3465339619346953 
11/26/03 12:00:00 AM EST 24932 5173 0.20748435745227017 
11/26/03 1:00:00 AM EST  20474 3860 0.1885317964247338 
11/26/03 2:00:00 AM EST  22456 3522 0.1568400427502672 
11/26/03 3:00:00 AM EST  24407 4268 0.17486786577621174 
11/26/03 4:00:00 AM EST  22573 5094 0.22566783325211537 
11/26/03 5:00:00 AM EST  27626 5725 0.2072323173821762 
11/26/03 6:00:00 AM EST  24091 5048 0.2095388319289361 
11/26/03 7:00:00 AM EST  17624 4336 0.2460281434407626 
11/26/03 8:00:00 AM EST  18059 3603 0.1995127083448696

This is actually quite good, normally my machine can only handle 10% of
its queries on a good day.  

sentData
The number of times data was sent, and whether sending was successful.
Aggregated over every minute . Recorded aggregates over hour:
11/26/03 3:00:00 AM EST 244 35 0.14344262295081966 
11/26/03 4:00:00 AM EST 262 43 0.16412213740458015 
11/26/03 5:00:00 AM EST 242 33 0.13636363636363635 
11/26/03 6:00:00 AM EST 205 35 0.17073170731707318 
11/26/03 7:00:00 AM EST 192 30 0.15625 
11/26/03 8:00:00 AM EST 177 30 0.1694915254237288
receivedData
The number of times data was received, and whether receiving was
successful. Aggregated over every minute . Recorded aggregates over
hour:
11/26/03 3:00:00 AM EST  97 56 0.5773195876288659 
11/26/03 4:00:00 AM EST  94 63 0.6702127659574468 
11/26/03 5:00:00 AM EST 141 65 0.46099290780141844 
11/26/03 6:00:00 AM EST  88 39 0.4431818181818182 
11/26/03 7:00:00 AM EST  77 32 0.4155844155844156 
11/26/03 8:00:00 AM EST  77 36 0.4675324675324675

Hard for me to say one way or another if these values are normal... I
seem to remember sentData having ugly percentages but not this low.
-------------------------------------------------------

This probably isn't worth thinking about as changing maxlog2data
probably introduces some bugs into the system, but some people in irc
seem to be curious about what would happen with these changes.

*** BTW, there was listed in the OCM for a short time last night an
inbound connection with 169.254.0.0 ... which afaik is an illegal
address (Windows 2K automatic ip addresser assigns it if it cant find a
dhcp server in certain configurations.)

Cheers :-)

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

Reply via email to