Hello everybody,
after upgrading my system from sarge to etch this weekend, "named" crashes frequently enough to be annoying leading to a reproducible bug. On my system processes are allowed to create coredumps. -rw------- 1 bind bind 26505216 2007-09-09 08:27 core.14257 -rw------- 1 bind bind 26505216 2007-09-09 08:31 core.15809 -rw------- 1 bind bind 26501120 2007-09-09 08:33 core.16422 -rw------- 1 bind bind 26378240 2007-09-09 08:35 core.16557 -rw------- 1 bind bind 26378240 2007-09-09 08:40 core.17151 -rw------- 1 bind bind 26378240 2007-09-09 08:55 core.18819 -rw------- 1 bind bind 26443776 2007-09-09 09:21 core.20212 -rw------- 1 bind bind 26517504 2007-09-08 17:00 core.24074 -rw------- 1 bind bind 26521600 2007-09-08 17:05 core.24840 -rw------- 1 bind bind 26521600 2007-09-08 17:12 core.25418 -rw------- 1 bind bind 26947584 2007-09-08 16:16 core.3466 -rw------- 1 bind bind 26521600 2007-09-08 16:20 core.6878 -rw------- 1 bind bind 26734592 2007-09-08 19:57 core.7026 -rw------- 1 bind bind 26742784 2007-09-09 07:51 core.7033 -rw------- 1 bind bind 26517504 2007-09-08 16:58 core.8949 A gdb session on the 20212 coredump reveals this backtrace: (gdb) bt full #0 0xb7f49410 in ?? () No symbol table info available. #1 0xb6a7509c in ?? () No symbol table info available. #2 0x00000006 in ?? () No symbol table info available. #3 0x00004ef7 in ?? () No symbol table info available. #4 0xb7b24811 in raise () from /lib/tls/i686/cmov/libc.so.6 No symbol table info available. #5 0xb7b25fb9 in abort () from /lib/tls/i686/cmov/libc.so.6 No symbol table info available. #6 0x08064a88 in ns_main_earlyfatal () No symbol table info available. #7 0xb7c858fb in isc_socket_detach () from /usr/lib/libisc.so.11 No symbol table info available. #8 0xb7c8706c in isc_socket_detach () from /usr/lib/libisc.so.11 No symbol table info available. #9 0xb7c873f9 in isc_socket_detach () from /usr/lib/libisc.so.11 No symbol table info available. #10 0xb7c32240 in start_thread () from /lib/tls/i686/cmov/libpthread.so.0 No symbol table info available. #11 0xb7bc74ae in clone () from /lib/tls/i686/cmov/libc.so.6 No symbol table info available. Right now the only workaround on this bug is to restart the died process using "monit" but I'll dig further into it. It seems to have something to do with the forward zones declared in my named.conf.local. They are all forwarded to systems reachable via VPN. On some occasions when the VPN goes down for a short period of time, named pulls 100% CPU and crashes with the error message shown in the subject after about 30 seconds. t++ -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
