What Eric said. With the changes in http2 worker scheduling, if I introduced a bug in child exit there, I'd expect it to trigger via workers_pool_cleanup() in h2_workers.c
-Stefan > Am 18.04.2017 um 14:47 schrieb Eric Covener <cove...@gmail.com>: > > On Tue, Apr 18, 2017 at 8:43 AM, Stefan Priebe - Profihost AG > <s.pri...@profihost.ag> wrote: >> bt of such a process shows: >> (gdb) bt >> #0 0x00007f5df74f64db in pthread_join () from >> /lib/x86_64-linux-gnu/libpthread.so.0 > > Need to see the other threads in the process, this one is just waiting > for the others to complete. > > "thread apply all bt full" > > > -- > Eric Covener > cove...@gmail.com