The dump and reboot process is significantly faster than in
previous releases of HP-UX, and can handle technologies such as
more than 4GB of physical memory, dynamically loaded kernel modules
(DLKM), non-contiguous physical memory, and the like.
System crash dumps do not contain the entire contents of physical memory
by default. With memory sizes growing in leaps and bounds, it is critical
that only those parts of physical memory which are considered useful
in debugging a problem are dumped. For a discussion of which parts
of memory are dumped, see “Selective Dumps”.
To simplify access of crash dumps by commands and utilities,
HP-UX provides a crash dump access library, libcrash,
which can be used to access any past-, current-, or future-format
dump. For more information, see “Debugging of Crash Dumps”.
The system crash dump user interface affords an operator sitting
at the console a limited degree of control of the dump process,
plus a great deal more information about its progress while dumping
than was available in the past. For more information, see “Dump-Time User Interface”.
Aside from speeding up the dump process itself, HP-UX speeds
up the process of saving the dump into the file system after the
system has rebooted. Proper use of the savecrash(1M) and crashutil(1M) utilities, combined with proper configuration of
dump devices, can greatly reduce or nearly eliminate the time that
used to be spent running savecore (now obsolete) in prior releases. For more information,
see “Post-Reboot Dump Processing”.
It is not necessary to rebuild the kernel in order to configure
dump devices: there are several ways of configuring dump devices.
Also described here are the ways to override the defaults for which
parts of memory get included in the dump. For more information,
see “Crash Dump Configuration”.
Printable version
