Testing Emule 0.60 (completed) Community version
#321
Posted 20 June 2021 - 10:40 AM
This is only binding the TCP port. The UDP ports (TCP+4 and kad port) are binding to 0.0.0.0
$ netstat -apn | grep -i wine
(Not all processes could be identified, non-owned process info
will not be shown, you would have to be root to see it all.)
tcp 2 0 10.16.128.127:6822 0.0.0.0:* LISTEN 1264795/wineserver6
udp 0 0 0.0.0.0:6826 0.0.0.0:* 1264795/wineserver6
udp 134912 0 0.0.0.0:6832 0.0.0.0:* 1264795/wineserver6
The significance of this is that if the host has multiple default route metrics the program will likely select the lowest one and send using the "wrong" IP
#322
Posted 20 June 2021 - 12:46 PM
it would be great if you answer to the question above: Does it differ from 0.50a?
stoatwblr,
external IP is unknown when a socket is created; especially in case of multiple network adapters.
#323
Posted 21 June 2021 - 04:40 PM
fox88, on 20 June 2021 - 01:46 PM, said:
external IP is unknown when a socket is created; especially in case of multiple network adapters.
You are correct, however if the bindaddr directive is being used then the TCP/UDP and Kad UDP ports should _all_ use them.
The directive specifically exists for the purpose of allowing manual control of the IP address the program may otherwise use or listen on
(there are good reasons why you may want to bind the webUI port to other interfaces, such as for internal network control etc, so it's best left listening on 0.0.0.0 or have a separate directive, as amule does)
#324
Posted 22 June 2021 - 12:24 PM
#325
Posted 22 June 2021 - 01:22 PM
fox88, on 22 June 2021 - 01:24 PM, said:
It's no longer built for windows (*nix/macos only). I'll check with some older binaries and get back to you
...
This is using a 2.3.2 contributed binary
netstat -apn | grep :46[0-9][0-9] | grep wine
tcp 0 0 10.172.65.13:4662 0.0.0.0:* LISTEN 1152762/wineserver6
udp 0 0 0.0.0.0:4665 0.0.0.0:* 1152762/wineserver6
udp 0 0 0.0.0.0:4672 0.0.0.0:* 1152762/wineserver6
So it looks like this is a wine/windows issue
Additionally, TCPdump shows no UDP traffic on the "unwanted" IP addresses(*), so it looks to be a false alarm
(*) The test host has one main address using nm-team and a bunch of macvlan attachments behind it with real world IPv4 addresses (doing this gives each interface its own mac address which simple IP aliasing doesn't do - that makes firewalling much easier)
This post has been edited by stoatwblr: 22 June 2021 - 01:39 PM
#326
Posted 22 June 2021 - 01:49 PM
I've verified (side by side comparions of connections) that MorphXT users still comprise a good chunk of the ed2k cloud and are blacklisting all emule 0.6* that they see
It's going to be virtually impossible to get people to update their MorphXT builds even if notices are posted on all the usual forums
As such, in order to prevent continued network fragmentation: I strongly urge renaming to 0.7 and skipping any further use of the 0.6 name
#327
Posted 04 August 2021 - 11:14 AM
* (quite old it's in 0.5* too): When importing downloads, the location of the last import isn't remembered, so one must navigate to the directory containing the partfiles each time
* (new): "enable import parts" only activates on a per-session basis (it's not saved)
* (old - and relatively major): sending to some clients (particularly amule) stalls out and fails before uploading a single chunk. I thought this was a new interaction but I've found reports of the behaviour dating back to 2011 and was originally reported when sending to shereaza clients. I've confirmed it with Amule, Shereaza and Neoloader. Things work fine in the other direction (this has some speed/latency issues going on. It doesn't happen if emule is uploading to the max number of clients (ie: throttling individual uploads), but does if the client in question is receiving at "high" speeds - the breakpoint is related to the upload speed limit setting). Above 200kB/sec it almost always happens, but at 1MB/s or faster several chunks may suceed before it breaks
* handling of long paths and some filename characters (eg: "?") is broken. These can appear on network shares
* Suboptimal filelist/directory scanning: It would be nice to discuss optimisations for sharing "lots of files" - emule can take up to 90 minutes to start/reload in some circumstances where amule sharing the same files does it in under 90 seconds...
This post has been edited by stoatwblr: 04 August 2021 - 11:21 AM
#328
Posted 04 August 2021 - 11:35 AM
around 2/3 of the network is running low IDs and at least 90% of those actually downloading files are lowID users with traffic routed via servers
This level of LowID is breaking things fairly badly
(*) I've had it happen when uPNP failed or the uPNP gateway was reset and occasionally on restarting emule when the bound port wasn't released from the last run, so the new session couldn't attach to the designated port (if this happens, the port will be released eventually by the OS)
#329
Posted 05 August 2021 - 04:38 PM
stoatwblr, on 04 August 2021 - 02:14 PM, said:
A bug is either unintended or unexpected behaviour.
Otherwise this should go as a feature request.
Also, if alternative clients in eMule's network have communication issues, this usually hints at a problem in these clients.
This post has been edited by fox88: 05 August 2021 - 04:39 PM
#330
Posted 05 August 2021 - 09:14 PM
fox88, on 05 August 2021 - 05:38 PM, said:
That's a bit like German car makers saying that they don't need to deal with issues "because that should never happen" - vs Japanese makers taking the stance "It does happen, so we need to deal with it"
Or in other words: "If architects designed like coders, the first woodpecker to come along would destroy civilisation"
#331
Posted 07 August 2021 - 02:48 PM
If you keep ignoring low fuel warning, sooner or later an out of gas issue might happen.
But that is not German car maker's fault.
#332
Posted 14 August 2021 - 03:24 PM
fox88, on 07 August 2021 - 03:48 PM, said:
If you keep ignoring low fuel warning, sooner or later an out of gas issue might happen.
But that is not German car maker's fault.
The primary reason for software failure is failure to interoperate or play nice with others
That's why Usenet and SMTP email worked so well whilst numerous other attempts failed
The same reason is why we have "plugfests" in the networking world - to ensure that every maker's equipment talks to every other maker's equipment. Just because a "standard" has been written, doesn't mean that it can't be interpreted in various ways and it isn't uncommon to find devices which are fully "standards compliant" but won't talk to each other due to differing approaches to that standard
Failure to address software interoperation is harmful to the ed2k/kad concept and will ensure that users continue to migrate to other software. Couple this with the serious networking issues (over 50% LowID, over 90% of my traffic is going via the ed2k servers to/from lowid clients) and there are major points of failure developing in what is supposed to be a robust distributed network
Interoperabilty has been a keystone of every sucessful protocol on the Internet. Brushing this issue off as "It's their fault" doesn't fly when these OTHER pieces of software all happily talk to and exchange data with each other. Emule is the odd one out, not them.
#333
Posted 02 September 2021 - 12:48 PM
#334
Posted 08 September 2021 - 09:42 AM
#335
Posted 24 November 2021 - 01:07 PM
It features minor fixes and improvements.
The link is in the topmost post.
#336
Posted 24 November 2021 - 01:24 PM
ed2k://|file|eMule0.60d_x64.zip|4628841|AAB171A72B92F6786F00DC0BA291E7B9|h=J76SRB4C7U7FFDME3J267LZ5UY3YOEHC|/
ed2k://|server|91.208.162.87|4232|/
ed2k://|server|85.239.33.123|4232|/
ed2k://|server|91.208.162.55|4232|/
SD - Telegram
Air VPN - The air to breathe the real Internet
BTC
bc1qdrk0ld07jtg99ym2zg68cpqhqj34qnf2txm93n
XMR
48ja6xJ2NyPMNzmY1pA3ZZPpX5yTaw9Ym28jrDPCL7Y7L7pr5wXFdpeK4WqBbvVY5qEa6VDfhFKTnHWef3EPC4zgQNTnAwg
#337
Posted 25 November 2021 - 05:43 AM
#339
Posted 27 November 2021 - 08:12 AM
#340
Posted 01 December 2021 - 09:54 AM
fox88, on 24 November 2021 - 11:07 PM, said:
It features minor fixes and improvements.
The link is in the topmost post.
Will it break anything to do so?Will it cause grief?