View Single Post
  #94 (permalink)  
Old September 28th, 2017
Lord of the Rings's Avatar
Lord of the Rings Lord of the Rings is offline
ContraBanned
 
Join Date: June 30th, 2004
Location: Middle of the ocean apparently (middle earth)
Posts: 973
Lord of the Rings has a distinguished reputationLord of the Rings has a distinguished reputationLord of the Rings has a distinguished reputation
Default WSHR/LW5ís upload logic is very buggy

WSHR/LW5ís upload logic is bugged. Since early in year Iíve been using fibre instead of ADSL.

(i) If I increase upload slots beyond 5, the upload speeds become chaotic. So chaotic that downloaders often drop off. By chaotic I mean the speeds rapidly fluctuate between zero to high speeds every second or two. This appears triggered when there actually are more than 5 uploads. So this is definitely a bug & not anything to do with my bandwidth capabilities after all.

WireShare  (formerly entitled LimeWire Pirate Edition)-chaotic-upspeeds-7-up-slots-2017-09-28.gif
5 snapshots over 19 seconds. Not ideal to show how chaotic it actually is. Each up-speed change occurs at a rate of 0.5 to 2 secs. Also this example does not show uploads dropping off. When that happens itís the ones at zero up-speed. Ideally I should have screen video-recorded this over 3 to 5 seconds. This occasion had up slots set to 7. Generally the up-speed bandwidth changes are much more dramatic, such as from 0 to 120 to 30 to 0, etc. within just a couple of seconds. I had only just started WSHR on this occasion after a router reboot.

(ii) Another side effect of the upload logic bug(s) is if there is a greedy host. That then causes upload bandwidth to vanish into overhead. ie: The total upload bandwidth will show maximum allocated but what the downloaders are able to utilise is a fraction of that. It hampers the speeds available to the downloaders, slowing them down. This generally lasts between 45 to 90 minutes before slowly going back to normal.

WireShare  (formerly entitled LimeWire Pirate Edition)-ws-upload-bugs-waiting.png (-from 2015-) WireShare  (formerly entitled LimeWire Pirate Edition)-ws-upload-bandwidth-bug.png

WireShare  (formerly entitled LimeWire Pirate Edition)-upload-bug-speed-when-no-one-downlding-2017-08-11.png (-from 2017-) WireShare  (formerly entitled LimeWire Pirate Edition)-buggered-upload-speed-due-greedy-host-2017-08-26.png

The same issue reported July 29th, 2011 here LimeWire Pirate Edition post titled: 'The Down-sides of LimeWire 5 / LPE - the super-bugs !' from paragraph starting with ďThe total of upload bandwidth at times either leaks or giving incorrect figures. ÖĒ onward. It also seems to affect the queuing process, ie: queued downloads that should have timed into starting to download are left in the queue for longer periods. (Ignore dates on the images, due to a forum crash & backup corruption needed to be re-uploaded in 2014. Original upload bug images dated 12 Dec 2011 & 6 April 2012.)

(iii) If you recall, the other upload bug I was able to overcome when I increased both the individual & overall queueing totals. That upload bug was generally initiated by a greedy host. Then other hosts whom only wanted to download a single file would either have their download stopped or refused & be auto-banned for an hour or longer (this happened to hosts of all modern clients.) ie: the program's auto-banning of a greedy host would spill across & affect one or more other download hosts.
This change most likely also fixed the 'Invalid Request' bug. Something I have very rarely seen since.

I donít know which is worse, WSHRís upload bugs or Phexís 2+ slot bug. If set Phex to more than 1 slot it becomes unlimited. I certainly donít want 50+ uploads at once. Start of this decade I told Phex dev about this but he seemed disinterested in fixing it. From memory LW4ís upload controls worked flawlessly & gave the sharer much more control over their uploads. (I used LW4 for 7 years after skipping LW3) Whilst I said flawlessly, there could always be the issue of a host stealing most of the bandwidth but that can happen with LW5/WSHR (there are a number of reasons this might happen, not just more hops or bad hops relaying to a specific host.)
Reply With Quote