On Fri, Sep 28, 2001 at 12:21:19AM +0100, Rory Campbell-Lange wrote: > I have been asked by an application developer, who is trying to find out > problems with one of their products running on our server, to set up the > server so that "core files can be written". What are these, and how would I > do this?
When a program crashes the last thing that happens is that its memory image is written to a file called core. Some shells allow to specify how big such core files may be; bash uses "ulimit -f <#kilobytes>". Default the max size is 0, so practically prohibiting core dumps. > I have also been instructed to, once a core file is generated, to "rename > the core file to a unique name and collect a stack backtrace". > > 1. How will I know when a core file has been generated? when the program crashed lookout for a file called core in your home dir or in the dir the program started from. -- groetjes, carel
