[2.0.1] Repeated SIGABRT before finish syncing #3844
Comments
NicolasGoeddel
changed the title
|
FYI @ogoffart The only assert here I see is |
|
This is also the same assert that @rmoesbergen is seeing in #3816 (comment) |
|
I've spent some time looking at this and unfortunately don't see what could cause this error. The assert seems correct, and the counting was correct (pushed a small cleanup). Everything should be happening in the same thread. Maybe this slot is called twice for some reason. But then MAX_PARALLEL=1 shouldn't help. @ogoffart Any ideas? |
|
Are _current and total volatile? Did you use a barrier to wait for the other threads to finish? |
|
@NicolasGoeddel The slot is always executed by the same thread. Can you compile the client yourself? If so, I'd love to give you some debugging code. |
|
Maybe I can try this at weekend because the client runs in a productive environment. Compiling should be no problem. It seems there are some Ubuntu packages missing which were not mentioned in the building instructions at https://doc.owncloud.org/desktop/2.0/building.htm |
|
After installing the packages "build-essential qt4-make qt4-dev-tools pkg-config libneon27-dev qtkeychain-dev libsqlite3-dev" I was able to compile the whole thing. But I do not understand why I should create the directory client-build and the run cmake ../client in it. After compiling the directory client-build remains empty and all binaries were put into client/bin . I would recommend to rework the compiling instructions. Can I start the newly compiled owncloud-client as a second instance or do I need to quit the running instance first? |
|
@NicolasGoeddel Thanks for trying this! You can run your selfmade client with --confdir path (create the directory first) to ensure it uses a different configuration from your production client. You have to quit your other client before starting this one though. |
|
@ckamm Ok then. If I have to quit the other client before starting the compiled one, then we have to wait until the day after tomorrow. It is important that the client runs flawlessly on this specific machine. It would not help to test your debugging code on an other machine. |
|
@NicolasGoeddel Okay then. The code is uploaded to my repo as e3b65c5. If you can compile it and run with --logfile my.log and then show all lines that contain "BUG!", that should help us figure out what's going on. |
|
@ckamm Please can you give me advice how to checkout this code? Or have I simply to do a 'git pull' and then recompile it with 'cmake -DCMAKE_BUILD_TYPE="Debug" .'? |
|
@NicolasGoeddel Something like this should do it: You do not need to compile in debug mode to get the log output. |
|
I am now trying to crash owncloud for half an hour and it don't want to crash. The log has already a size of 1.4 GB. I hope it won't grow too fast. |
|
Any news here? |
|
@NicolasGoeddel So it crashed easily with "2.0.1, Revision 58ea49" but you could not yet make "ckamm/debug" crash? |
|
I was able to crash it after a week again. But this time it's an other error and my log has a size of 44 GB. I can try to grep all lines of the log with a "BUG!" in it into a new file if that would help. But I don't think so. |
|
@NicolasGoeddel and me talked on IRC. The "bad_alloc" crash was most likely due to logging being enabled. He'll try to reproduce the problem again with --logdir. |
ghost
modified the milestones:
|
@NicolasGoeddel Did this clear up? |
|
The owncloud client is running since Oct 2 without problems. The latest log file is "owncloud.log.27398". I am still using the version "ckamm/debug". So I guess the problem is gone. I don't know why. I could check the old version again, if you think that would help. |
|
@NicolasGoeddel The debug patch didn't change any logic, just add debug output. Maybe you could try with a regular build again? |
|
Yeah. I've just got a log with some BUG!s in it. The main problem seems to be this line: I would like to send you the tar.gziped logfile per mail because I don't want to make the whole directory structure available here. |
|
I got the log and will take a deeper look tomorrow. A first glance shows that We should fix the bug in that job and ideally make this robust against bugs in jobs. |
|
@NicolasGoeddel Fixed in master, thanks for the log file! :) |
ckamm
added
ReadyToTest
|
@ckamm Nice! I will check out the fix the next days and test it. |
|
Closing for now :-) You know when/how to re-open.. |
Hi there,
since a few days I have a problem with my Owncloud-Client 2.0.1. It starts, begins to sync and then it gives me a SIGABRT before finish syncing.
Client: Version 2.0.1, Revision 58ea49
OS: Lubuntu 14.04.3 (3.16.0-46-generic x86_64)
Server: Ubuntu 14.04.3
Apache: Version 2.4.7
SSL: Yes
This is the backtrace from gdb. The issue seems to occur in the main Thread.
At the moment I solved the problem presumably by exporting
OWNCLOUD_MAX_PARALLEL=1before starting owncloud.I also have a log file but it has a few hundred MB. I don't know why it grows so fast.
The text was updated successfully, but these errors were encountered: