Description of problem: It should be possible to enable coredump capture in the appliance. This would allow us to more easily debug certain problems without requiring 'printf' statements as now. It would work something like this: (1) Add a handle variable called 'coredump' / LIBGUESTFS_COREDUMP=<path> (2) This sets a kernel parameter for the appliance. (3) If the kernel parameter is set then the daemon enables coredumps: (a) echo /sysroot/<path>/core.%t.%e > /proc/sys/kernel/core_pattern (b) Set ulimit core unlimited (4) After something coredumps (even the daemon) this ought to capture the coredump on the currently mounted device. Version-Release number of selected component (if applicable): 1.5.x
I wonder if we can coredump to a raw disk? Then the user could add a "coredump capture disk".
Unless we plan to turn this on by default, this is really a developer tool. I think we can assume to a significant extent that the user understands exactly how it works and what the implications of using it are. This would include, for example, choosing a sensible value for LIBGUESTFS_COREDUMP for the current image file. I'd keep this one simple.
Patch posted upstream: https://www.redhat.com/archives/libguestfs/2010-July/msg00090.html
Setting back to ASSIGNED. The patch was not acceptable upstream. This requires a fix to the kernel, which the kernel developers have rejected.