aMule Forum
English => aMule crashes => Topic started by: guest on September 18, 2010, 08:19:49 PM
-
aMule (both amule and amuled) is crashing after few seconds since the last Ubuntu's kernel update :(
Here is the backtrace... It's running on a 64 bit platform (Ubuntu 10.04) and IS NOT a development version.
----------------------------=| BACKTRACE FOLLOWS: |=----------------------------
Current version is: aMuled 2.2.6 using wxGTK2 v2.8.10
Running on: Linux 2.6.32-24-generic x86_64
[2] ?? in amuled[0x42de34]
[3] wxFatalSignalHandler in /usr/lib/libwx_baseu-2.8.so.0[0x7f1db81ae41c]
[4] ?? in /lib/libpthread.so.0[0x7f1db955b8f0]
[5] wcslen in /lib/libc.so.6[0x7f1db761e130]
[6] wxString::Replace(wchar_t const*, wchar_t const*, bool) in /usr/lib/libwx_baseu-2.8.so.0[0x7f1db81724b7]
[7] ?? in amuled[0x53e276]
[8] ?? in amuled[0x467197]
[9] ?? in amuled[0x46df4f]
[10] ?? in amuled[0x439823]
[11] ?? in amuled[0x42b232]
[12] wxEntry(int&, wchar_t**) in /usr/lib/libwx_baseu-2.8.so.0[0x7f1db8150452]
[13] ?? in amuled[0x42a722]
[14] __libc_start_main in /lib/libc.so.6[0x7f1db75aec4d]
[15] ?? in amuled[0x429cc9]
--------------------------------------------------------------------------------
Aborted
Bye.
-
Did you have aMule from Ubuntu's repository?
Can anybody with 10.04 64bit latest update reproduce this?
-
Dunno if I have the 64bit version, but if so I'll post a follow-up on this.
-
Did you have aMule from Ubuntu's repository?
yes (italian version)
-
There is no "Italian version" of aMule.
Are you possibly talking about aMule Adunanza ?
-
Maybe its totally unrelated (as its about GPU lockups),
but there are some problems with Ubuntu 10.4 regarding Intel chipsets (https://wiki.ubuntu.com/X/Bugs/Lucidi8xxFreezes).
So, do you have an Intel chipset?
If so, you might consider going back to 9.10, as its supported until April 2011,
and 10.10 is released in October.
-
No, I have 3 AMD processors, 2.1 GHz each one.
-
There is no "Italian version" of aMule.
Are you possibly talking about aMule Adunanza ?
guest, kindly answer this question.
-
There is no "Italian version" of aMule.
Are you possibly talking about aMule Adunanza ?
It's aMule, not aMule Adunanza. I was talking about my Ubuntu language....
-
Fine, then.
-
I don't know why, but I solved removing the "out of disk space" event... Now it goes :)
-
Were you out of disc space?
-
Yes, but it crashed only if an event was assigned...
I'm using these scripts: http://wiki.amule.org/index.php/Events#Linux (http://wiki.amule.org/index.php/Events#Linux).
-
I didn't get it to crash when it runs out of disc space with events. However I tried the SVN version in Maverick.
-
Different guest here - running x64 Ubuntu 10.10 on AMD dual core with 4 gb ram and 2.5 gb swap. After about 30 mins Amule starts using all avaliable ram and swap until the system crashes or I have to force a reset. Run a Nvidia 7950 card on an AMD motherboard (inbuilt raedon graphics) so no intel chips to confuse the issues. I am not to sure what is causing it but it certainly a memory issue.
-
What version of aMule do you have?
-
What is your kernel version? Did you update to the latest kernel?
-
Was using 2.2.6 release - Tried downloading the latest and rolling my own but I'm hitting upnp problems when trying to compile using 10400. I haven't been able to resolve whats causing this as yet. Kernel version 2.6.35-23 generic amd 64. Currently posting this on -22 and will try aMule to see if it crashes on this as well.
-
Kernel version 2.6.35-22 generic amd 64 works ok. Compiked 10400 so will try that this weekend on -23
-
Ran it for a week on -22 - no issues. Ran 10400 on -23 and ran into every expanding memory error within 2 hours. Unable to do a backtrace as waiting for the ~1-2 hours for the HDD to stop thrashing (swap) is not me. I use the button on the front.
-
Please run a top in parallel to see if the memory runs up gradually or in a quick surge.
If it's gradually, maybe you can run it with valgrind's memcheck to see where it leaks?
A backtrace is useless anyway in an out of memory condition.
-
Runs up slowly to start with then accelerates in a burst - got a link to running a top as I have never done that before.
-
Just open a console and run top :D
Equally simple question - how do I roll back to an older kernel (and then forward to a newer one) ?
-
Equally simple question - how do I roll back to an older kernel (and then forward to a newer one) ?
In Ubuntu you can do that easily using StartUp-Manager - see here:
http://ubuntuforums.org/showthread.php?t=1247937
-
I have the choice of the last couple of kernals when booting (at grub stage) just in case the newer one is cack for some programs or just cack, so I use it on occasions. With regards top should I pipe it to a file. Usually once the mem problem happens I cannot regain control of the system as i/o is flooded and hogs all resources.
-
OK. Can you try a few things to help track the problem down:
- Try latest SVN. I have plugged a memory leak in 10417, though I doubt it was The Big One.
- Try amuled with amulegui instead of amule, to see if it makes a difference. If it crashes - which of the two uses up the memory?
- Try amule but disable Kad, use only ED2K.
-
I'll run the latest SVN for a couple of days and see what happens on kernel 24 :)
By the way, I've written about the problem on the Ubuntu forums - even if there's no reaction from Ubuntu developers, at least the aMule users who go there to look for a solution will be able to find it.
-
Ran the release build on -24 and it chews up the memory but slower and is non-responsive (i/o bound) after a while. Going to compile the latest -svn soon and try it out
-
OK, I've been running SVN 10424 on the -24 kernel for a few days now and there were exactly zero crashes in that time. HOWEVER, I only have a few small (a couple hundred megabytes each) and rare (less than ten sources) files in my download queue. I'm going to add a huge and very popular file now and we'll see what happens :)
-
Well, when I came back home my GNOME was dead which means that the leak is still present. The file I've started downloading is humongous (almost 8 gigabytes) and very popular (200 sources, my max number) and I'm 100% sure that it is causing the problem. I've seen exactly the same behavior in the past when I tried to download huge and popular files.
I've turned off Kad now - let's see what happens :)
-
Thank you for helping track down this nasty problem. :)
Please try also to compile wxWidgets 2.8.11 yourself and use THAT to build aMule.
-
Thank you for helping track down this nasty problem. :)
Please try also to compile wxWidgets 2.8.11 yourself and use THAT to build aMule.
I did - I followed the guide in the Wiki and it said to compile wxWidgets first (wxGTK-2.8.11 to be precise) :)
-
Backtrace posted
http://forum.amule.org/index.php?topic=18538.0
-
My GNOME crashed yet again, so turning off Kad does not fix the problem. It does seem to make it happen less frequently, but I'm not 100% sure.
When I have a little more time I'll read up on using amuled with amuleGUI and I'll test that :)
-
Everybody please try this patch if it helps:
Index: EMSocket.cpp
===================================================================
--- EMSocket.cpp (revision 10426)
+++ EMSocket.cpp (working copy)
@@ -466,7 +466,7 @@
//printf("* Disconnected socket %p\n", this);
SocketSentBytes returnVal = { false, 0, 0 };
return returnVal;
- } else if (m_bBusy && onlyAllowedToSendControlPacket) {
+ } else if (m_bBusy) {
//printf("* Busy socket %p\n", this);
SocketSentBytes returnVal = { true, 0, 0 };
return returnVal;
-
You have to tell me how to use that thing ;)
-
Edit src/EMSocket.cpp line 469 from old (-) to new (+) and compile it.
-
soak testing it now
-
backtrace posted in same thread as before
-
Tried running valgrind on amule but after two hours amule hadn't come up. One of my cpus was pretty much maxed for the two hours so I'm guessing my comp might be a bit light on to run both. Might try it overnight tonight.
-
Unfortunately, the patch didn't help - it only took my aMule about an hour to crash.
-
Umm, Olhado, may you try to disable speed limits and see if amule keeps crashing?
-
I've disabled speed limits and aMule has been running for more than two hours now. Kad is enabled and there are large and popular files in the download queue. I probably won't be able to test it more thoroughly today since it makes browsing almost impossible, but it looks good so far :)
-
Good to see, I've done quite a bit of testing and in 100% of cases the crashes and OOM were occuring only with limits enabled.
If you are in linux you may use something like wondershaper in order to run amule without limits. Other workaround is to pause enough downloads so your download speed is reasonable.
Note that I'm just having the problem with the download speed limit, I'm perfectly fine using an upload limit.
-
My aMule has been running non-stop since the last time I've posted. That's quite a few hours and I'm pretty much certain that a crash should have occurred in that time. I've turned on the upload limit and we'll see what happens but it seems that you're right about the download limit causing the problem :)
-
Another backtrace:
Current version is: aMule 2.2.6 using wxGTK2 v2.8.11
Running on: Linux 2.6.35-24-generic i686
[2] ?? in amule[0x8080399]
[3] wxFatalSignalHandler in /usr/lib/libwx_baseu-2.8.so.0[0x6c5896]
[4] ?? in [0xfe6400]
[5] __pthread_mutex_lock in /lib/libpthread.so.0[0x84a3bd]
[6] ?? in /lib/libglib-2.0.so.0[0x4d1e118]
[7] g_io_add_watch_full in /lib/libglib-2.0.so.0[0x4d159b0]
[8] gdk_input_add_full in /usr/lib/libgdk-x11-2.0.so.0[0x780d71]
[9] gdk_input_add in /usr/lib/libgdk-x11-2.0.so.0[0x780dbe]
[10] GSocketGUIFunctionsTableConcrete::Install_Callback(GSocket*, GSocketEvent) in /usr/lib/libwx_gtk2u_core-2.8.so.0[0x37df85]
[11] GSocket::Enable(GSocketEvent) in /usr/lib/libwx_baseu_net-2.8.so.0[0x82716d]
[12] GSocket::Write(char const*, int) in /usr/lib/libwx_baseu_net-2.8.so.0[0x82853a]
[13] wxSocketBase::_Write(void const*, unsigned int) in /usr/lib/libwx_baseu_net-2.8.so.0[0x8237d6]
[14] wxSocketBase::Write(void const*, unsigned int) in /usr/lib/libwx_baseu_net-2.8.so.0[0x823947]
[15] ?? in amule[0x81fbb6d]
[16] ?? in amule[0x80de56c]
[17] ?? in amule[0x80dd621]
[18] ?? in amule[0x80b1424]
[19] ?? in amule[0x8120d23]
[20] wxThreadInternal::PthreadStart(wxThread*) in /usr/lib/libwx_baseu-2.8.so.0[0x6bffb8]
[21] wxPthreadStart in /usr/lib/libwx_baseu-2.8.so.0[0x6c001d]
[22] ?? in /lib/libpthread.so.0[0x847cc9]
[23] clone in /lib/libc.so.6[0xee669e]
Crashes in Ubuntu systematilly. The solution does not limit download is good temporarily, but it is only provisional. My system is updated to the latest versions of official ubuntu's kernel. Sorry for my english, i'am spanish. Thanks :)
-
Huh?
Your amule version is 2.2.6...
-
I guest, did this crash occur when limiting download?
-
I'm the guest witht eh amd x64 system. Ran amule overnight no problems crashing while limits turned off. Whereas everytime I crashed before I had limits enabled. I also did notice that the download speed was an issue as it throttled itself to less than the upload speed. If I enabled some limits the ratio was much better. Would the changes that occured in the kernal be affecting the socket/connection handling of amule?
-
SVN 10429 on amd x64 ubuntu -24 kernal crashes if download limits are enabled. If they are disabled but enabled at a later stage it crashes quite quickly.
-
Ok, fits into the picture. Now all we need is a solution. :(
-
Hi all,
I've been having the same problem.I'm very glad I found this post because setting the download limit to 0 has solved the issue for me. I thought it was the 2.2.6 aMule version from the RPMfusion repos, but I compiled SVN 10443 with all the options and had the same problem.
aMule has been running crash free for 11 hours now which is about 10.5 hours longer than it has ever been running recently.
Just for info, my o/s is Fedora 14 (x86_64) , kernel 2.6.35.10-74. I saw somewhere in the posts here that this issue may have something to do with glib, so its version is: Glib2 2.26.0-2.fc14.
Also, I saw a question posed as to the timing of the RAM and swap space getting used up. In my case, everything would be running very stable, and then all of a sudden, the RAM filled, and then swap space started filling very fast (within a second or two) which caused system lockup.
I hope that this info helps track the issue. Feel free to contact me for help or more info. I compiled with debug so I could do testing if someone tells me what they want.
best regards :)
-
Setting the download limit to 0 doesn't really solve the issue on my side. Instead it seems to limit the probability of occuring. Ever since the OOM issue arised I kept the download limit at 0 and even so it still crashes from time to time. On much fewer occasions, but still it does.
-
Setting the download limit to 0 doesn't really solve the issue on my side. Instead it seems to limit the probability of occuring. Ever since the OOM issue arised I kept the download limit at 0 and even so it still crashes from time to time. On much fewer occasions, but still it does.
AFAICT it could make sense, but I couldn't reproduce this exact crash with DL set to 0. Does the backtrace look the same?
^marcell^ what kind of load is in your eMule? Number of files, network capacity, average and maximum dl speed, upload speed, number of sources, etc.. Knowing that would be useful as then I'd try to replicate the load and the crash with DL set to 0.
Of course you may send me the date by PM if you are not confortable.
-
It probably also depends on the behavior of the other clients, like if they will close the connection if you download too slow or not. Which makes it quite random.
-
I've made a lot of testing and I never had trouble when the dl limit was set to 0, but it seems marcell has a new bug.
IMHO is very important to understand if dl limiting is the key factor for the bug.
-
How do I obtian a backtrace if the OS is killing aMule, because it allocates too much memory?
I am going to start aMule right now (using gdb) and hope you can give me an answer so I can post more information.
-
How do I obtian a backtrace if the OS is killing aMule, because it allocates too much memory?
marcell see the other thread (in amule bugs), there my script is posted (it would be good to read it all)
Anyways what I do is to monitor amule and when it passes 3Gb of memory I fire up gdb, then I get all the backtraces of amule eating memory that I want.
-
Hi all, Kry posted a wx patch in the other thread (look in bugs forum) which seems to solve the issue for me.
Just a rebuilt of wx is needed, I encourage you to try it.
-
It seems that the patch solved the issue for me too :) I have three very popular files in my download queue, my download limit is set to 10 kb/s, and aMule has been running all day with zero problems.
-
Ah, good news. :D