On Mon, 2 Mar 2009, Evan Layton wrote: > Alok Aggarwal wrote: >> Thinking about this some more, it seems to me >> that the fix for 6556 should be coordinated with >> the fix for 4407 otherwise we run the risk of potentially not being able to >> debug a bad install. > > That's true unless we do this fix such that if the install fails or hits an > error we don't do the auto-reboot.
Jan's proposal accounts for the auto-reboot to not be done in case of a failed install. But, there are cases (bugs in the code, etc) where a failure might not be propagated appropriately. The logs are our only source of information in that case after the reboot has been done (defect 4407). Alok
