I've been writing a daemon in node (hosted by forever) and notice that the worker process sometimes randomly dies. I added a segault handler to see what's up, but the typical stacktrace is... unhelpful?
PID 9949 received SIGSEGV for address: 0x0
/home/wesley/app/node_modules/segfault-handler/build/Release/segfault-handler.node(+0xf60)[0x7f11dd3d8f60]
/lib64/libpthread.so.0[0x354760f710]
[0x3d3c10b0d799]
Once I received this trace on failure, instead:
PID 5820 received SIGSEGV for address: 0x18
/home/wesley/app/node_modules/segfault-handler/build/Release/segfault-handler.node(+0xf60)[0x7fdfab749f60]
/lib64/libpthread.so.0[0x354760f710]
/usr/bin/node(_ZN4node6Buffer14MakeFastBufferERKN2v89ArgumentsE+0x168)[0x426778]
[0x1d06e36ccdfb]
Which still shows the top call in libpthread.so (Though preceded by a call in node proper.)
I find it difficult to believe that the segfault originates in libpthread.so (or... nowhere?), but that's what it looks like? Has anyone had any experiences like this in the past? Does anyone have any tips for further debugging?
I'm running on CentOS 6.5 with node v0.10.29 (from the epel repository), for reference.