Jump to content

2.2 - MTU Discovery does not work . Seeding does not reach # of slots


Recommended Posts

-- 2010-09-10: Version 2.2 Beta (build 21882)

- Feature: end to end path MTU calculation based on ICMP messages and missing packets

- Defining 10 UL slots, reaching 4-5 slots :(

- MTU is fixed/"stuck" @ 1444 bytes (like 2.04)

- DF flags are all set to defragment

See for yourselves:


capture: http://www.mediafire.com/?zrvwcqc3cexhjbf

Did anyone really tested seeding with this release ? ... Checked that all back-ported features (from 3.0) really do work ? :(

Back to 2.04 . Let me know when it's tested and working. You know my email... :P


And if you did test it on the latest 3.0 and it DOES work there- let me know as well. Thank you.

Link to comment
Share on other sites

Just discussed that with alus on IRC... He'll have to clean up (delete) all the settings for that #. Pref->bandwidth, properties, setup guide ect... Maybe just display it somewhere .

I guess just writing the code w/o modifying the related UI - is not enough... :P

Link to comment
Share on other sites

Yes (as I understood - it's all global and automatic now). Though personally I would like to have some global control over it, it IS one less item to bother the simple user with...

May be a global setting with default - 0 (=auto) can be best.

btw, people do report on problems of low UL speed in the 2.2 thread... so I guess alus might have still some tuning to do :P

Link to comment
Share on other sites

Whether it's in normal settings or advanced settings, I don't care.

But I do care if it actually works.

I'm seeing too many upload connections running at <0.5 KB/sec, both from distant sources or even sometimes from my end. This is often caused by uTorrent's behavior rather than ISP throttling.

Torrent priority probably should affect it -- no point giving a torrent set to "LOW" a lot of upload slots while one set to "HIGH" also has lots of peers you could upload to.

Link to comment
Share on other sites

Well, we can discuss/re-test seeding after the UI/setting for the new chocker will be revised to set a global max.

On the main issue - PMTUD - here is another test:

- My MTU was set to 1392 (on my PC/NICs)

- seeding 1 torrent in uTP/ V2.2

-uploading to 3-5 peers (mostly v2.04 peers)

expected result - un-fragmented packets of size a bit below my set MTU

Result: packet size reaches 1402 (1410) and the DF flag is not set, showing - that packets are still being fragmented (my guess: 1402 + 42 = 1444 ...)

On some peers - the size did drop to ~762 (not-frequented). But - 2 packets of 700 + 700 are like 1400 + 40 as far as PPS goes....

So, I guess they should all be decremented and by smaller steps/decrements. When you did this simple sanity test, what were your results ?

screens: http://img841.imageshack.us/g/42821809.png/

capture: http://www.mediafire.com/?4llfbuzxurdyu1y

My conclusion - it is broken...

Link to comment
Share on other sites

  • 4 weeks later...
  • 2 weeks later...

One more test:

This one starts with a single torrent with 5K/10K/20K/25K/30K upload limit. It continues for ~10 hours with the 25K limit. Using uTP + TCP.

On the way - RSS added two more downloads, that had about 2.hours download and 1.5 hours seeding each ... ;)



Full log (16M, sorry, 400M unzipped):


Bottom lines:

1. # of slots are (with 1 torrent):

- ~9 @ 5K

- ~8 @ 10K

- ~11 @ 20K

- ~8 @ 25K

- ~4 @ 30K

Peers-connections are having 0.6-3K speed when speed is <25K - too low speed/slot , and too many slots/UL speed.

I think there should be a set limit of >4-5K/slot and >3-4 slots / torrent. Also 0 the actual UL limit should be taken into account!

2. The is a strange drop every 5 minutes, with retransmit overhead. You should find the cause of this (not sure if it is the choker itself)

3. Totals are good here:

1.03G uploaded, 2.18G (= 3.22G) downloaded - Data pnly

1.1G uploaded , 2.3G (= 3.50G) downloaded - Data+ overhead (traffic cap)

~10% overhead

So, tuning 1 , and finding the cause for 2 will do nicely for a release of a good and easy to use choker :)

Other tests @ a connection with higher speed are desired.

Link to comment
Share on other sites


This topic is now archived and is closed to further replies.

  • Create New...