Force bind to do core dump

Juan Bernhard juan at inti.gob.ar
Mon May 16 18:58:19 UTC 2016


Hello list, I have this problem, if some one can give me a hand...
	Im using bind910 (from ports) and freebsd 10.3-RELEASE-p2 on 2 dns 
servers, both in production. From time to time, bind gets a "pid 13762 
(named), uid 53: exited on signal 6"
	Im trying to report this problem to ISC, but I cant get to named do a 
core dump... every time I kill -6 `pgrep named` its never do a core dump 
(on any freebsd I have). I even run bind from root to troubleshoot any 
file permission...
This is the process information:

# procstat -f `pgrep named`
   PID COMM               FD T V FLAGS     REF  OFFSET PRO NAME
78816 named             text v r r-------  -       - - 
/usr/local/sbin/named
78816 named              cwd v d r-------  -       - - 
/usr/local/etc/namedb/zonas
78816 named             root v d r-------  -       - -   /
78816 named                0 v c rw------  4       0 -   /dev/null
78816 named                1 v c rw------  4       0 -   /dev/null
78816 named                2 v c rw------  4       0 -   /dev/null
78816 named                3 s - rw------  1       0 UDD /var/run/logpriv
78816 named                4 v c rw------  4       0 -   /dev/null
78816 named                5 p - rw---n--  2       0 -   -
78816 named                6 v r -wa-----  1 150810810 -   -
78816 named                7 p - rw------  1       0 -   -
78816 named                8 k - rw------  2       0 -   -
78816 named                9 v c r----n--  1    1656 -   /dev/random
78816 named               20 s - rw---n--  2       0 ?
[I deleted the IP sockets FD, they don't matter to this]

# procstat -l `pgrep named`
   PID COMM             RLIMIT                  SOFT             HARD
78816 named            cputime             infinity         infinity
78816 named            filesize            infinity         infinity
78816 named            datasize               32768 MB         32768 MB
78816 named            stacksize             524288 KB        524288 KB
78816 named            coredumpsize        infinity         infinity
78816 named            memoryuse           infinity         infinity
78816 named            memorylocked        infinity         infinity
78816 named            maxprocesses            6670             6670
78816 named            openfiles              58275            58275
78816 named            sbsize              infinity         infinity
78816 named            vmemoryuse          infinity         infinity
78816 named            pseudo-terminals    infinity         infinity
78816 named            swapuse             infinity         infinity

# sysctl -a | grep core
kern.corefile: %N.core
kern.nodump_coredump: 0
kern.coredump: 1
kern.capmode_coredump: 0
kern.sugid_coredump: 0
kern.coredump_pack_vmmapinfo: 1
kern.coredump_pack_fileinfo: 1
debug.ncores: 5
debug.elf32_legacy_coredump: 0
debug.elf64_legacy_coredump: 0


coredumpsize is infinity, so this is not a problem, the CWD is writeable 
by the process (but it is open as read only, maybe this is the 
problem?). This SHOULD give me a core dump in 
/usr/local/etc/namedb/zonas/named.core but it never appear.
According to ISC, bind should give a core dump on signal 6
https://kb.isc.org/article/AA-00340/0/What-to-do-if-your-BIND-or-DHCP-server-has-crashed.html

I tried to kill -6 a vi process and the system DOES a core dump. But not 
from bind

Im about to send this email to ISC also, but I would like to argue thats 
not freebsd problem.

Thank you.

Saludos, Juan.





More information about the freebsd-questions mailing list