That is probably a know issue, caused not by a distribution nor a thread
library but by ThreadedDataSource design:
You will find suggested temporary solution there as well. To get a final
one whole ThreadedDataSource concept needs to be rethinked.
Diana Cionoiu napisa?(a):
> Hello Etoile,
> Which distro do you use?
> Etoile Dièse wrote:
>> We recently installed Yate (version 1.0.0) on a machine wich takes
>> between 2000 and 4000 incoming calls per day (not very heavy loaded).
>> The problem is that Yate cores quite often (something like once an
>> hour) and it is catastrophic for us.
>> Yate is mainly used for IVR with a Digium card (Zaptel driver). We
>> heavily use conference and wavefile modules. We feel (only a feeling)
>> that the problem comes from the interruptions of play file when
>> pressing a key. The cores say that the SIGSEGV is received on "delete
>> tmp;" line 874, in file engine/DataFormat.cpp, method
>> ThreadedSource::stop called by Wavefile, line 176 in file
>> modules/wavefile.cpp, method WaveSource::~WaveSource.
>> It seems that the wavefile module corrupts something wich leads to
>> that core later. Optionaly, I would like to know if those messages
>> from Yate are normals and, if not, how to avoid them :
>> ThreadPrivate terminating pthread 0x585c90 [0x585c80]