Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Gomaaz]

hmm no probs here  ?( pretty strange your problem

[ken]

Mars_Artis, could you do a couple of things that may help us figure out what's going on?

When aMule stops downloading and gets low-id, don't quit it.  Instead, please do the following:

Important: for all the following steps, replace "4662" with your aMule TCP port number.

[*]Run the Terminal program in /Applications/Utilities
[*]In Terminal, type this command: netstat -aLp tcp | grep -w 4662
[*]Then type this: netstat -ap tcp | grep "\*\.4662"
[*]Then type this: telnet localhost 4662
The telnet program will either report an error or it will tell you it has connected.  If it connects, it will tell you what the "escape character" is.  It will probably be "^]".  That means Ctrl-].  Type that character and then type "quit" at the "telnet> " prompt.
[*]Copy the output from those commands and post it to this thread.
[*]Now, you can quit and restart aMule if you like.  You can also quit Terminal.
[/list]

Thanks for your help.

[Matreco]

I have the same problem...
When i start Amule i get high id for about 2 to 3 hours then i get disconnected and when i reconnect i always get low id...
This is the results i get in the terminal:

Last login: ************** on console
Welcome to Darwin!
bl6-67-152:~ ******$ netstat -aLp tcp | grep -w 4662
8/8/5          *.4662                
bl6-67-152:~ ******$ netstat -ap tcp | grep "\*\.4662"
tcp4       0      0  *.4662                 *.*                    LISTEN
bl6-67-152:~ ******$ telnet localhost 4662
Trying ::1...
telnet: connect to address ::1: Connection refused
Trying 127.0.0.1...
telnet: connect to address 127.0.0.1: Operation timed out
telnet: Unable to connect to remote host
bl6-67-152:~ ******$

[lionel77]

guys, i have a feeling that the problem is not restricted to 10.4.1.
i've installed tiger a couple days ago and now i am experiencing the same problem under 10.4.0.

right now, amule (2.0.3) has been running for 20h and is still uploading and connected to razorback. but when i try to connect to its tcp port through the connection tester i get a firewalled warning. in a couple of hours, upload will break down as well.

netstat & telnet give me exactly the same output that matreco has posted.

i'll do some more testing...

[Mars_Artis]

Ok mates here it's what i get from terminal:

Mars-G5:~ mars$ netstat -aLp tcp | grep -w 17672
8/8/5          *.17672                
Mars-G5:~ mars$ netstat -ap tcp | grep "\*\.17672"
tcp4       0      0  *.17672                *.*                    LISTEN
Mars-G5:~ mars$ telnet localhost 17672
Trying ::1...
telnet: connect to address ::1: Connection refused
Trying 127.0.0.1...
telnet: connect to address 127.0.0.1: Operation timed out
telnet: Unable to connect to remote host
Mars-G5:~ mars$


For me too making the port test as above gives firewalled error
I'll stay tuned for infos.

P.S.
I wanan add this: i did try to downgrade to Tiger 10.4.0 before the up to 10.4.2 And i did notice the some issue. No benefit at all

[ken]

OK, thanks for the reports, guys.  Unfortunately, there's no smoking gun.

The next step, I think, is for me to put diagnostic print-out statements throughout the wxMac and aMule code which deals with listening for and accepting connections.  I'll supply patches to lionel77, and he and I will run some tests.

I'm also open to suggestions from the gallery.  

[Gomaaz]

Can't believe it ! The prob just happened again with 10.4.2

God help us all  ;(

[lionel77]

i'm "glad" to hear that -- everybody else i know is experiencing the issue and it was really puzzling to me why the two of you apparently did not.

btw, you can track the accumulation of stuck connections with the "netstat -aLp tcp | grep -w 4662" command. right after you started amule the output should be "0/0/5 *.4662". then, as connections get stuck the first two numbers will increase successively. eventually you will get "8/8/5 *.4662", which is when the problem becomes visible in terms of low ids from servers and in terms of reduced download performance and eventually upload performance.

ken and i, did some extensive testing and we are pretty confident that the problem is located below amule and wx. out of curiosity, i installed mldonkey again to test whether it is suffering from the same limitations and it is. mldonkey is compiled with a maximum of 20 stuck connnections instead of 8, so it takes a bit longer until you can actually see the problem without using netstat, but it is definitely there.

so what's next? ken is trying to devise a small program/script that demonstrates the issue of stuck connections, so we are able to file a precise bug report with apple. there is also a chance that apple is already aware of the issue since it is not unlikely that other types of programs beside p2p apps are affected by this issue as well. so if we are lucky, 10.4.3 might solve our problems...

if everything else fails i am thinking about providing people with a modified amule setup: i've compiled wx to provide amule with a queue of 20 connections instead of 8, which delays the onset of the problem. to avoid getting stuck, i have written a small script that shuts down amule regularly via amulecmd and then restarts it. it's not a very clean solution but it might be better than nothing until apple fixes their sockets.

[Gomaaz]

hi first of all,

thank you for spending time in our problem

but I think the shutdown and restart script isnt a good solution because  emule network needs time to connect, to stay in queues and so on .

so we'll have to be depend on apple  for a real fix?

[lionel77]

i fully agree. restarting amule every couple hours is not a good solution. this is really just an emergency plan.

however, i am relatively confident that you don't lose your queue position if you restart amule and connect to a source again in time. the only problem is that you might be downloading a number of popular files with many sources, then you won't be able to reconnect to all sources in time to maintain your queue positions.

i hope we get this thing fixed soon so we don't need plan b. i still remember running mldonkey using a similar script before the devs fixed a bug that caused problems with os x 10.2. with the constant restarts mldonkey worked somehow, but it wasn't that great.

[Gomaaz]

heres a better way:

Put in the tiger installation disk
- boot from cd
- reinstall tiger (in the options you can click on "save users and network settings"
- reupdate to the newest version of tiger

after restarting amule seems to work fine now. All in all it spends a time about 40 mins.

but amule works....for a while  :]

[Gomaaz]

amule is quitting after  30mins automatically

I'll shutdown amule now for a while  

[lionel77]

Gomaaz:
I don't think reinstalling tiger does anything. And most importantly, if you want to test whether the stuck connections problem is present or not you really need to use "netstat -aLp tcp | grep -w 4662" as described above. Just observing amule is not enough.
I'm very sure that the problem was always present on your machine since you first updated to tiger, even at those times when you thought it wasn't. For some reason the rate of connections getting stuck might have been lower at times so that the problem was there latently but you didn't experience it b/c the socket was not completely stuffed.

[pedrito]

So, did anyone came up with a solution other than reverting to a previous version of the OS? I too have the same problem with Mac OS X 10.4.2 and aMule 2.0.3: It connects in an instant with High ID, performs nicely for several hours, the it looses connection and upone reconnecting all I got is a Low ID. And running "netstat -aLp tcp | grep -w 4662" also returns  "8/8/5          *.4662" (both when connected with a High ID and Low ID, though), whatever that means.

Is someone NOT suffering from this with the same software set up?

[lionel77]

A quick update:
It looks like there are a few people who accumulate stuck connections at a very low rate (1 per 2 days or so). My guess is that this is because of their particular setup (e.g. router, ISP or something along those lines).

I have patched wx now to allow for 193 stuck connections instead of the default 8. This delays the onset of the stuck connection effect tremendously. I've had different test versions running here without any apparent problems because of this modification.

I am in the process of preparing a binary to be posted here. However, I want to make sure that this version works well (both on 10.3 and on 10.4), so I need a few days more of testing.