.

setting the core dump name schema

Update: this is also a very useful defence against a currently known severe security affecting bug. See Core Dump Handling Vulnerability for information and workarounds

Lets assume you got a couple of core dumps files in one directory (e.g. home) and you don't know which program caused it.

Unless you have a grsec/pax kernel with logging features enabled, it might get tricky to find out who caused the coredumps.

This is the point where the naming scheme comes handy :)

The pattern can be read/set via /proc/sys/kernel/core_pattern.

To set a new one just do:

echo "newpattern" > /proc/sys/kernel/core_pattern

you can have variables to make the file named different per executable, pid a.s.o..

max length 64 characters; default value is "core"

Here is a small list of possible variables

%p:       pid
%<NUL>:   '%' is dropped
%%:       output one '%'
%u:       uid
%g:       gid
%s:       signal number
%t:       UNIX time of dump
%h:       hostname
%e:       executable filename
%<OTHER>: both are dropped

If core_pattern does not include "%p" (default does not) and core_uses_pid is set, then .PID will be appended to the filename.

example:

echo "core.%e.%p" > /proc/sys/kernel/core_pattern

core.<executable>.<pid>

to make the changes permanent, add the following line to /etc/sysctl.conf:

kernel.core_pattern = core.%e.%p

happy core dumping.