kexec-tools/kdump.init

371 lines
8.5 KiB
Bash

#! /bin/sh
#
# kdump
#
# Description: The kdump init script provides the support necessary for
# loading a kdump kernel into memory at system bootup time,
# and for copying away a vmcore at system panic time.
#
# Copyright 2005 Red Hat, Inc.
#
# chkconfig: - 20 80
#
# Author: Jeff Moyer <jmoyer@redhat.com>
# Source function library.
. /etc/init.d/functions
KEXEC=/sbin/kexec
# Will be different for ia64, for example. For now, that architecture isn't
# supported. Code needs to be added here when we do.
BOOTDIR="/boot"
KDUMP_KERNELVER=""
KDUMP_COMMANDLINE=""
KEXEC_ARGS=""
KDUMP_CONFIG_FILE="/etc/kdump.conf"
LOGGER="/usr/bin/logger -p info -t kdump"
standard_kexec_args="-p"
if [ -f /etc/sysconfig/kdump ]; then
. /etc/sysconfig/kdump
fi
function save_core()
{
coredir="/var/crash/`date +"%Y-%m-%d-%H:%M"`"
mkdir -p $coredir
cp /proc/vmcore $coredir/vmcore-incomplete
if [ $? == 0 ]; then
mv $coredir/vmcore-incomplete $coredir/vmcore
$LOGGER "saved a vmcore to $coredir"
else
$LOGGER "failed to save a vmcore to $coredir"
fi
}
function check_config()
{
if [ -z "$KDUMP_KERNELVER" ]; then
local running_kernel=`uname -r`
kdump_kver=`echo $running_kernel | sed 's/smp//g'`
else
kdump_kver=$KDUMP_KERNELVER
fi
kdump_kernel="${KDUMP_BOOTDIR}/${KDUMP_IMG}-${kdump_kver}${KDUMP_IMG_EXT}"
kdump_initrd="${KDUMP_BOOTDIR}/initrd-${kdump_kver}kdump.img"
if [ ! -f $kdump_kernel ]; then
echo -n "No kdump kernel image found."; warning; echo
echo "Tried to locate ${kdump_kernel}"
return 0
fi
if [ ! -f $kdump_initrd ]; then
echo -n "No kdump initial ramdisk found."; warning; echo
echo "Rebuilding $kdump_initrd"
/sbin/mkdumprd -d -f $kdump_initrd $kdump_kver
if [ $? != 0 ]; then
echo "Failed to run mkdumprd"
$LOGGER "mkdumprd: failed to make kdump initrd"
exit 1
fi
return 0
fi
#check to see if config file has been modified after
#last build of the image file
config_time=0
if [ -f $KDUMP_CONFIG_FILE ]; then
config_time=`stat -c "%Y" $KDUMP_CONFIG_FILE`
fi
kernel_time=`stat -c "%Y" $kdump_kernel`
image_time=`stat -c "%Y" $kdump_initrd`
if [ "$config_time" -gt "$image_time" ] ||
[ "$kernel_time" -gt "$image_time" ]; then
echo "Detected $KDUMP_CONFIG_FILE or $kdump_kernel change"
echo "Rebuilding $kdump_initrd"
/sbin/mkdumprd -d -f $kdump_initrd $kdump_kver
if [ $? != 0 ]; then
echo "Failed to run mkdumprd"
$LOGGER "mkdumprd: failed to make kdump initrd"
return 1
fi
fi
}
# This function check iomem and determines if we have more than
# 4GB of ram available. Returns 1 if we do, 0 if we dont
function need_64bit_headers()
{
MEMSZ=`sed -e's/\(^[0-9]\+.*\)/0x\1/' -e's/\(-\)\(.*\)/ 0x\2/' /proc/iomem | \
awk 'BEGIN {
totalmem=0;
segmentmem=0;
}
/.*RAM.*/ {
start = strtonum($1);
end = strtonum($2);
segmentmem=end-start;
totalmem=totalmem+(segmentmem/1024);
if (end >= 4194304) {
totalmem=totalmem+4194304;
}
}
END {
printf "%d", (totalmem+1);
}'`
#The AWK script above computes the total number of KB
#in the RAM areas of /proc/iomem
#Note also that we set totalmem to 4GB in the event that
#any physical address is larger than 4GB. This lets us
#default to 64 bit ELF headers for PAE kernels, which
#need then to access those higher addresses.
#This comparison tells us if the above amount is more than
#4GB (4096 KB). I do the funny math to avoid overflow
if [ $MEMSZ -ge 4194304 ]
then
return 1
fi
return 0
}
# Load the kdump kerel specified in /etc/sysconfig/kdump
# If none is specified, try to load a kdump kernel with the same version
# as the currently running kernel.
function load_kdump()
{
if [ -z "$KDUMP_COMMANDLINE" ]
then
KDUMP_COMMANDLINE=`cat /proc/cmdline`
fi
MEM_RESERVED=`echo $KDUMP_COMMANDLINE | grep "crashkernel=[0-9]\+[MmKkGg]@[0-9]\+[MmGgKk]"`
if [ -z "$MEM_RESERVED" ]
then
$LOGGER "No crashkernel parameter specified for running kernel"
return 1
fi
ARCH=`uname -m`
if [ "$ARCH" == "i686" -o "$ARCH" == "i386" ]
then
need_64bit_headers
if [ $? == 1 ]
then
FOUND_ELF_ARGS=`echo $KEXEC_ARGS | grep elf32-core-headers`
if [ -n "$FOUND_ELF_ARGS" ]
then
echo -n "Warning: elf32-core-headers overrides correct elf64 setting"
warning
echo
else
KEXEC_ARGS="$KEXEC_ARGS --elf64-core-headers"
fi
else
FOUND_ELF_ARGS=`echo $KEXEC_ARGS | grep elf64-core-headers`
if [ -z "$FOUND_ELF_ARGS" ]
then
KEXEC_ARGS="$KEXEC_ARGS --elf32-core-headers"
fi
fi
fi
KDUMP_COMMANDLINE=`echo $KDUMP_COMMANDLINE | sed -e 's/crashkernel=[0-9]\+[MmKkGg]@[0-9]\+[MmGgKk]//'`
KDUMP_COMMANDLINE="${KDUMP_COMMANDLINE} ${KDUMP_COMMANDLINE_APPEND}"
$KEXEC $KEXEC_ARGS $standard_kexec_args \
--command-line="$KDUMP_COMMANDLINE" \
--initrd=$kdump_initrd $kdump_kernel 2>/dev/null
if [ $? == 0 ]; then
$LOGGER "kexec: loaded kdump kernel"
return 0
else
$LOGGER "kexec: failed to load kdump kernel"
return 1
fi
}
function propagate_ssh_key()
{
#Check if selinux is on... must flip to permissive mode
#for the moment to create key, then flip back...
se_enforce=`/usr/sbin/sestatus | grep -c "^Current mode.*enforcing"`
if [ "$se_enforce" -ge 1 ]; then
/usr/sbin/setenforce 0 2>&1 > /dev/null
fi
#Use dedicated key for kdump
if [ ! -d /root/.ssh ]; then
mkdir /root/.ssh
chmod 700 /root/.ssh
fi
if [ ! -e /root/.ssh/config ]; then
echo "IdentityFile ~/.ssh/kdump_id_rsa" >> /root/.ssh/config
elif [ `grep -c kdump_id_rsa /root/.ssh/config` -eq 0 ]; then
echo "IdentityFile ~/.ssh/kdump_id_rsa" >> /root/.ssh/config
fi
local KEYFILE=/root/.ssh/kdump_id_rsa.pub
local errmsg="Failed to propagate ssh key"
#make sure they've configured kdump.conf for ssh dumps
local SSH_TARGET=`awk '/^\ *net.*@.*$/ {print $0}' $KDUMP_CONFIG_FILE`
if [ -z "$SSH_TARGET" ]; then
echo "No ssh config specified in $KDUMP_CONFIG_FILE. Can't propagate"
$LOGGER "$errmsg, no ssh config specified in $KDUMP_CONFIG_FILE"
exit 1
fi
#Check to see if we already created key, if not, create it.
if [ -f $KEYFILE ]; then
echo "Using existing keys..."
else
echo -n "Generating new ssh keys... "
/usr/bin/ssh-keygen -t rsa -f /root/.ssh/kdump_id_rsa -N "" 2>&1 > /dev/null
echo "done."
fi
#If necessary, flip selinux back to enforcing
if [ "$se_enforce" -ge 1 ]; then
/usr/sbin/setenforce 1 2>&1 > /dev/null
fi
#now find the target ssh user and server to contact.
SSH_USER=`echo $SSH_TARGET | cut -d\ -f2 | cut -d@ -f1`
SSH_SERVER=`echo $SSH_TARGET | sed -e's/\(.*@\)\(.*$\)/\2/'`
#now send the found key to the found server
cat $KEYFILE | ssh -x $SSH_USER@$SSH_SERVER "mkdir -p ~$SSH_USER/.ssh; chmod 700 ~$SSH_USER/.ssh; cat >> ~$SSH_USER/.ssh/authorized_keys2"
RET=$?
if [ $RET == 0 ]; then
echo $KEYFILE has been added to ~$SSH_USER/.ssh/authorized_keys2 on $SSH_SERVER
$LOGGER "propagated ssh key (ssh server: $SSH_SERVER)"
return 0
else
echo $KEYFILE failed in transfer to $SSH_SERVER
$LOGGER "$errmsg, unable to transfer $KEYFILE to $SSH_SERVER"
exit 1
fi
}
function status()
{
if [ ! -e /sys/kernel/kexec_crash_loaded ]
then
return 2
fi
rc=`cat /sys/kernel/kexec_crash_loaded`
if [ $rc == 1 ]; then
return 0
else
return 1
fi
}
function start()
{
#TODO check raw partition for core dump image
status
rc=$?
if [ $rc == 2 ]; then
echo -n "Kdump is not supported on this kernel"; failure; echo
return 1;
else
if [ $rc == 0 ]; then
echo -n "Kdump already running"; success; echo
return 0
fi
fi
check_config
if [ $? != 0 ]; then
echo -n "Starting kdump:"; failure; echo
$LOGGER "failed to start up, config file incorrect"
return 1
fi
load_kdump
if [ $? != 0 ]; then
echo -n "Starting kdump:"; failure; echo
$LOGGER "failed to start up"
return 1
fi
echo -n "Starting kdump:"; success; echo
$LOGGER "started up"
}
function stop()
{
$KEXEC -p -u 2>/dev/null
if [ $? == 0 ]; then
$LOGGER "kexec: unloaded kdump kernel"
echo -n "Stopping kdump:"; success; echo
$LOGGER "stopped"
else
$LOGGER "kexec: failed to unload kdump kernel"
echo -n "Stopping kdump:"; failure; echo
$LOGGER "failed to stop"
fi
}
case "$1" in
start)
if [ -s /proc/vmcore ]; then
save_core
reboot
else
start
fi
;;
stop)
stop
;;
status)
EXIT_CODE=0
status
case "$?" in
0)
echo "Kdump is operational"
;;
1)
echo "Kdump is not operational"
EXIT_CODE=1
;;
2)
echo "Kdump is unsupported on this kernel"
EXIT_CODE=1
;;
esac
exit $EXIT_CODE
;;
restart)
stop
start
;;
condrestart)
;;
propagate)
propagate_ssh_key
;;
*)
echo $"Usage: $0 {start|stop|status|restart|propagate}"
exit 1
esac
exit $?