kdump-config - configure kdump, check status, or save a vmcore
file
kdump-config { help | test | show | status | load | unload |
savecore | propagate | symlinks {kernel-version} }
kdump-config manages the kdump feature of the Linux kernel.
When a kdump enabled kernel panics, it immediately boots into a clean kernel
called the kdump kernel. The memory image of the panicked kernel will be
presented in /proc/vmcore while the kdump kernel (or "capture
kernel") is running.
kdump-config loads the kdump kernel into the current kernel
and after a kdump, captures the vmcore image via kdump kernel.
kdump-config can take advantage of the linux-image-$(uname
-r)-dbg package and will try to use makedumpfile to capture the
vmcore. The crash tool can then be used to analyze the vmcore at a
later time.
kdump-config exits with 0 on success and 1 on failure.
- help
- Print a usage summary.
- test
- Determine the parameters that would be used to load the kdump kernel, but
do not load the kdump kernel. Useful for evaluating your configuration
settings.
- show
- Show current parameters and the saved kexec command from the last
kdump-config load.
- status
- Evaluate /sys/kernel/kexec_crash_loaded or
/sys/kernel/fadump_register depending on the dump mode and print a
corresponding message.
- load
- Load the kdump kernel or register if using firmware assisted dump
(fadump)
- unload
- Unload the kdump kernel or unregister if using firmware assisted dump
(fadump)
- savecore
- If /proc/vmcore exists, save it using makedumpfile.
- propagate
- Sends public ssh key to remote host for passwordless connection
- symlinks
{kernel-version}
- Verify and create vmlinuz and initrd.img links in
/var/lib/kdump according to the provided kernel version. If the
target initrd.img file is absent, it will create it.
- /etc/init.d/kdump-tools
- init script to automatically load a kdump kernel, or save a vmcore and
reboot.
- /etc/default/kdump-tools
- configuration file
- /var/crash/kernel_link
- a link to the current debug kernel
- /var/crash/kexec_cmd
- the last kexec_cmd executed by kdump-config
- /var/lib/kdump/vmlinuz
- Symbolic link pointing to the current kernel file in /boot
- /var/lib/kdump/initrd.img
- Symbolic link pointing to the initrd.img file with the current
kernel version in /boot
- USE_KDUMP is not
set or zero
- The variable USE_KDUMP in the file /etc/default/kdump-tools
is 0. To use kdump, edit the file and change the value to 1.
- kdump is not supported by
this kernel
- The file /sys/kernel/kexec_crash_loaded does not exist. This means
that the currently running kernel either was not configured to support
kdump, or that a crashkernel= commandline parameter was not used when the
currently running kernel booted.
- no crashkernel= parameter in
the kernel cmdline
- a crashkernel= commandline parameter was not used when the currently
running kernel booted. You will need to add a crashkernel= parameter to
your /etc/lilo.conf, /etc/elilo.conf or grub
/boot/grub/menu.lst file and reboot, before kdump will work. See
installed READMEs for suggestions.
- KDUMP_KERNEL
does not exist: $KDUMP_KERNEL
- A KDUMP_KERNEL was specified in /etc/default/kdump-tools, but it
cannot be found.
- KDUMP_INITRD
does not exist: $KDUMP_INITRD
- A KDUMP_INITRD was specified in /etc/default/kdump-tools, but it
cannot be found.
no KDUMP_KERNEL in $KDUMP_DEFAULTS and
can't find a kdump package directory in /usr/lib
no kdump kernel could be found.
found a linux-kdump package in $KDUMP_PKG_DIR
but can't find a kdump kernel in $KDUMP_PKG_DIR
this indicates a broken linux-kdump package
installation.
no kdump initrd image: $KDUMP_INITRD
an initrd will not be used
this indicates that although a kdump kernel was found, a
corresponding initrd was not found.
- kdump kernel
relocation address does not match crashkernel= parameter
- For non-relocatable architectures, the kdump kernel must be built with a
predetermined start address. This message indicates that the start address
of the kdump kernel and the start address in the crashkernel= parameter do
not match.
- failed to load kdump
kernel
- kexec reported that it could not load the kdump kernel.
- failed to unload
kdump kernel
- kexec reported that it could not unload the kdump kernel. Run 'kexec -u'
directly for more detail.
- makedumpfile
failed, falling back to 'cp'
- this message is printed if the kdump-tools init script attempts to
save a /proc/vmcore file using makedumpfile and the save
fails.
- failed to save
vmcore in $KDUMP_STAMPDIR
- this message indicates that makedumpfile and the fallback method of using
cp has failed. This usually occurs if pathnames are wrong or if the disk
is full.
- Invalid argument :
missing kernel version
- The symlinks command was used with a kernel version that is not
installed on this server
- Invalid symlink :
{symbolic link}
- The listed symbolic link is absent, broken or pointing to the wrong
version
- Cannot change symbolic
links when kdump is loaded
- The defined symbolic links are already in use. You must unload kdump
before using the symlinks command
- Unable to locate kernel
hook
- The kdump-tools script responsible for the creation of smaller
initrd.img files is absent from /usr/kernel/postinst.d
- Invalid kernel
version : {kernel version}
- The kernel version given to the symlinks command does not
exist
- fadump is not
configured in this kernel.
- The Firmware Assisted Dump (fadump) facility is not available on this
platform
- SSH and NFS cannot be defined
concurrently
- Both variables are defined in /etc/default/kdump-tools. You must choose
one or the other as configuration option.
- Unable to mount
remote NFS directory {directory}. Cannot save core
- The remote NFS server does not authorize this server to mount the
directory to save the crash dump.
- Unable to write to
the remote NFS directory {directory}. Cannot save core
- The remote NFS directory can be mounted but this server cannot write files
to it.
- makedumpfile
--dump-dmesg failed. dmesg content will be unavailable
- The command to capture the content of the dmesg buffer failed. It will not
be available in the crash directory
- Unable to reach
remote server {server}. No reason to continue
- kdump-tools is configured for remote dump but the remote server
cannot be reached.
- makedumpfile
scp failed. The vmcore file will not be available
- The makedumpfile command was successful but the resulting file
could not be copied over SSH to the remote server.
- Remote ssh dump is not
configured. No reason to propagate"
- The propagate command was used but kdump is not configured for
remote dump.
Written by Terry Loftin <terry.loftin@hp.com> and Louis
Bouchard <louis.bouchard@ubuntu.com>