kexec-tools/kdump-lib-initramfs.sh
Tao Liu fc66e25f7b Re-introduce vmcore creation notification to kdump
Upstream: fedora
Resolves: RHEL-70214
Conflict: Yes, the conflict is the same as the original c9s commit
	  c5aa4609 ("Introduce vmcore creation notification to kdump")
	  9ec61f6c ("Return the correct exit code of rebuild initrd")
          Also this patch cherry-picked the ipv6 fixed in [1].

[1]: https://github.com/rhkdump/kdump-utils/pull/60/files

commit 24e76222c740def1d03a506652400fe55959e024
Author: Tao Liu <ltao@redhat.com>
Date:   Fri Nov 29 16:15:18 2024 +1300

    Re-introduce vmcore creation notification to kdump

    Motivation
    ==========

    People may forget to recheck to ensure kdump works, which as a result, a
    possibility of no vmcores generated after a real system crash. It is
    unexpected for kdump.

    It is highly recommended people to test kdump after any system modification,
    such as:

    a. after kernel patching or whole yum update, as it might break something
       on which kdump is dependent, maybe due to introduction of any new bug etc.
    b. after any change at hardware level, maybe storage, networking,
       firmware upgrading etc.
    c. after implementing any new application, like which involves 3rd party modules
       etc.

    Though these exceed the range of kdump, however a simple vmcore creation
    status notification is good to have for now.

    Design
    ======

    Kdump currently will check any relating files/fs/drivers modified before
    determine if initrd should rebuild when (re)start. A rebuild is an
    indicator of such modification, and kdump need to be tested. This will
    clear the vmcore creation status specified in $VMCORE_CREATION_STATUS,
    and as a result, a notification of vmcore creation test will be
    outputted.

    To test kdump, there is an entry for doing that by "kdumpctl test". It
    will generate a timestamp string as the ID of the current test, along
    with a "pending" status in $VMCORE_CREATION_STATUS, then a real crash &
    dump process will be triggered.

    After system reboot back to normal, a vmcore creation check will start at
    "kdumpctl (re)start/status", and will report the results as
    success/fail/manual status to users.

    To achieve that, program will first check the status in $VMCORE_CREATION_STATUS.
    If "pending" status if found, which means the test result is
    undetermined and need a retrive from remote/local dump folder. Then if test
    id is found in the dump folder and vmcore is complete, then "pending"
    would be overwritten by "success", which indicates a successful kdump
    test. If test id is found in the dump folder but vmcore is incomplete,
    then it is a "fail" kdump test. If no test id is found, then it is a "manual"
    status, which indicates users should check the test results manually.

    If $VMCORE_CREATION_STATUS is already success/fail/manual status, it indicates
    the test result has already been determined, so the program will not access
    the remote/local dump folder again. This can limite any unnecessary
    access to dump target, shorten the time consumption.

    User should check for the root cause of fail/manual status when get
    reports.

    $VMCORE_CREATION_STATUS is used for recording the vmcore creation status of
    the current env. The format is like:

       <status> kdump_test_id=<timestamp sec>-<timestamp nanosec>
    e.g:
       success kdump_test_id=1729823462-938751820

    Which means, there has been a successful kdump test at
    $(date -d "@1729823462") timestamp for the current env. Timestamp
    nanosec is only meaningful for uniquify id string.

    Difference
    ==========
    Previously there is one commit 88525ebf ("Introduce vmcore creation
    notification to kdump") merged and addressing the same issue, but
    implemented differently:

    The prev one:
    Save the $VMCORE_CREATION_STATUS to local drive during the 2nd kernel
    dumping. If vmcore dumping target is different from $VMCORE_CREATION_STATUS's
    drive, then the latter one need to be mounted in 2nd kernel.

    This one:
    Save the $VMCORE_CREATION_STATUS to local drive only in 1nd kernel, that
    is, the test result is retrived after 2nd kernel dumping. So it doesn't
    load or mount other drive in 2nd kernel.

    The advantage:
    Extra mounting in 2nd kernel will introduce higher risk of failure,
    as a result, lower the success of vmcore dumping, which is
    unaccepted. So keep the code for 2nd kernel as simple is preferred.

    Usage
    =====
    [root@localhost ~]# kdumpctl restart
    kdump: kexec: unloaded kdump kernel
    kdump: Stopping kdump: [OK]
    kdump: kexec: loaded kdump kernel
    kdump: Starting kdump: [OK]
    kdump: Notice: No vmcore creation test performed!

    [root@localhost ~]# kdumpctl status
    kdump: Kdump is operational
    kdump: Notice: No vmcore creation test performed!

    [root@localhost ~]# kdumpctl test

    [root@localhost ~]# cat /var/lib/kdump/vmcore-creation.status
    pending kdump_test_id=1729823462-938751820

    [root@localhost ~]# kdumpctl status
    kdump: Kdump is operational
    kdump: Notice: Last successful vmcore creation on Fri Oct 25 02:31:02 AM UTC 2024

    [root@localhost ~]# cat /var/lib/kdump/vmcore-creation.status
    success kdump_test_id=1729823462-938751820

    [root@localhost ~]# kdumpctl restart
    kdump: kexec: unloaded kdump kernel
    kdump: Stopping kdump: [OK]
    kdump: kexec: loaded kdump kernel
    kdump: Starting kdump: [OK]
    kdump: Notice: Last successful vmcore creation on Fri Oct 25 02:31:02 AM UTC 2024

    Note: the notification for kdumpctl (re)start/status can be disabled by
    setting VMCORE_CREATION_NOTIFICATION in /etc/sysconfig/kdump. And fadump
    is NOT supported for this feature.

    Signed-off-by: Tao Liu <ltao@redhat.com>

Signed-off-by: Tao Liu <ltao@redhat.com>
2024-12-06 15:27:20 +13:00

189 lines
4.2 KiB
Bash
Executable File

#!/bin/sh
#
# The code in this file will be used in initramfs environment, bash may
# not be the default shell. Any code added must be POSIX compliant.
DEFAULT_PATH="/var/crash/"
KDUMP_CONFIG_FILE="/etc/kdump.conf"
FENCE_KDUMP_CONFIG_FILE="/etc/sysconfig/fence_kdump"
FENCE_KDUMP_SEND="/usr/libexec/fence_kdump_send"
LVM_CONF="/etc/lvm/lvm.conf"
# Read kdump config in well formated style
kdump_read_conf()
{
# Following steps are applied in order: strip trailing comment, strip trailing space,
# strip heading space, match non-empty line, remove duplicated spaces between conf name and value
[ -f "$KDUMP_CONFIG_FILE" ] && sed -n -e "s/#.*//;s/\s*$//;s/^\s*//;s/\(\S\+\)\s*\(.*\)/\1 \2/p" $KDUMP_CONFIG_FILE
}
# Retrieves config value defined in kdump.conf
# $1: config name, sed regexp compatible
kdump_get_conf_val()
{
# For lines matching "^\s*$1\s+", remove matched part (config name including space),
# remove tailing comment, space, then store in hold space. Print out the hold buffer on last line.
[ -f "$KDUMP_CONFIG_FILE" ] &&
sed -n -e "/^\s*\($1\)\s\+/{s/^\s*\($1\)\s\+//;s/#.*//;s/\s*$//;h};\${x;p}" $KDUMP_CONFIG_FILE
}
is_mounted()
{
findmnt -k -n "$1" > /dev/null 2>&1
}
# $1: info type
# $2: mount source type
# $3: mount source
# $4: extra args
get_mount_info()
{
__kdump_mnt=$(findmnt -k -n -r -o "$1" "--$2" "$3" $4)
[ -z "$__kdump_mnt" ] && [ -e "/etc/fstab" ] && __kdump_mnt=$(findmnt -s -n -r -o "$1" "--$2" "$3" $4)
echo "$__kdump_mnt"
}
is_ipv6_address()
{
echo "$1" | grep -q ":"
}
is_fs_type_nfs()
{
[ "$1" = "nfs" ] || [ "$1" = "nfs4" ]
}
is_fs_type_virtiofs()
{
[ "$1" = "virtiofs" ]
}
# If $1 contains dracut_args "--mount", return <filesystem type>
get_dracut_args_fstype()
{
echo $1 | grep "\-\-mount" | sed "s/.*--mount .\(.*\)/\1/" | cut -d' ' -f3
}
# If $1 contains dracut_args "--mount", return <device>
get_dracut_args_target()
{
echo $1 | grep "\-\-mount" | sed "s/.*--mount .\(.*\)/\1/" | cut -d' ' -f1
}
get_save_path()
{
__kdump_path=$(kdump_get_conf_val path)
[ -z "$__kdump_path" ] && __kdump_path=$DEFAULT_PATH
# strip the duplicated "/"
echo "$__kdump_path" | tr -s /
}
get_root_fs_device()
{
findmnt -k -f -n -o SOURCE /
}
# Return the current underlying device of a path, ignore bind mounts
get_target_from_path()
{
__kdump_target=$(df "$1" 2> /dev/null | tail -1 | awk '{print $1}')
[ "$__kdump_target" = "/dev/root" ] && [ ! -e /dev/root ] && __kdump_target=$(get_root_fs_device)
echo "$__kdump_target"
}
get_fs_type_from_target()
{
get_mount_info FSTYPE source "$1" -f
}
get_mntpoint_from_target()
{
local _mntpoint
# get the first TARGET when SOURCE doesn't end with ].
# In most cases, a SOURCE ends with ] when fsroot or subvol exists.
_mntpoint=$(get_mount_info TARGET,SOURCE source "$1" | grep -v "\]$" | awk 'NR==1 { print $1 }')
# fallback to the old way when _mntpoint is empty.
[[ -n "$_mntpoint" ]] || _mntpoint=$(get_mount_info TARGET source "$1" -f )
echo $_mntpoint
}
is_ssh_dump_target()
{
kdump_get_conf_val ssh | grep -q @
}
is_raw_dump_target()
{
[ -n "$(kdump_get_conf_val raw)" ]
}
is_virtiofs_dump_target()
{
if [ -n "$(kdump_get_conf_val virtiofs)" ]; then
return 0
fi
if is_fs_type_virtiofs "$(get_dracut_args_fstype "$(kdump_get_conf_val dracut_args)")"; then
return 0
fi
if is_fs_type_virtiofs "$(get_fs_type_from_target "$(get_target_from_path "$(get_save_path)")")"; then
return 0
fi
return 1
}
is_nfs_dump_target()
{
if [ -n "$(kdump_get_conf_val nfs)" ]; then
return 0
fi
if is_fs_type_nfs "$(get_dracut_args_fstype "$(kdump_get_conf_val dracut_args)")"; then
return 0
fi
if is_fs_type_nfs "$(get_fs_type_from_target "$(get_target_from_path "$(get_save_path)")")"; then
return 0
fi
return 1
}
fs_dump_target()
{
kdump_get_conf_val "ext[234]\|xfs\|btrfs\|minix\|virtiofs"
}
is_fs_dump_target()
{
[ -n "$(fs_dump_target)" ]
}
is_lvm2_thinp_device()
{
_device_path=$1
_lvm2_thin_device=$(lvm lvs -S 'lv_layout=sparse && lv_layout=thin' \
--nosuffix --noheadings -o vg_name,lv_name "$_device_path" 2> /dev/null)
[ -n "$_lvm2_thin_device" ]
}
kdump_get_ip_route()
{
if ! _route=$(/sbin/ip -o route get to "$1" 2>&1); then
exit 1
fi
echo "$_route"
}
kdump_get_ip_route_field()
{
echo "$1" | sed -n -e "s/^.*\<$2\>\s\+\(\S\+\).*$/\1/p"
}