On Thu, Mar 01, 2018 at 02:17:17PM -0700, Andreas Dilger wrote:
> On Mar 1, 2018, at 11:23 AM, Darrick J. Wong <darrick.wong@xxxxxxxxxx> wrote:
> >
> > From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> >
> > Implement online fsck for ext* filesystems which live on LVM-managed
> > logical volumes. The basic strategy mirrors that of e2croncheck --
> > create a snapshot, fsck the snapshot, report whatever errors appear,
> > remove snapshot. Unlike e2croncheck, this utility accepts any LVM
> > device path, knows about snapshots running out of space, and can call
> > fstrim having validated that the fs metadata is ok.
>
> One high-level note - in my lvcheck script, there was an option to disable
> background checking if the system was running on a battery:
I don't think e2scrub or e2scrub_all should refuse to run if the system
is on AC power -- if the user runs them from the cli then they shouldn't
have to override that kind of decision to get what they asked for.
However, checking for AC power certainly makes sense for the background
scrubber. The systemd service description contains a predicate to
disable the service if the system is running on a battery. However, the
cron job does not, so I will add that.
> # determine whether the machine is on AC power
> function on_ac_power() {
> local any_known=no
>
> # try sysfs power class first
> if [ -d /sys/class/power_supply ]; then
> for psu in /sys/class/power_supply/*; do
> if [ -r "$psu/type" ]; then
> type=$(cat "$psu/type")
>
> # ignore batteries
> [ "$type" = "Battery" ] && continue
>
> online=$(cat "$psu/online")
>
> [ "$online" = 1 ] && return 0
> [ "$online" = 0 ] && any_known=yes
> fi
> done
>
> [ "$any_known" = "yes" ] && return 1
> fi
>
> # else fall back to AC adapters in /proc
> if [ -d /proc/acpi/ac_adapter ]; then
> for ac in /proc/acpi/ac_adapter/*; do
> if [ -r "$ac/state" ]; then
> grep -q on-line "$ac/state" && return 0
> grep -q off-line "$ac/state" && any_known=yes
> elif [ -r "$ac/status" ]; then
> grep -q on-line "$ac/status" && return 0
> grep -q off-line "$ac/status" && any_known=yes
> fi
> done
>
> [ "$any_known" = "yes" ] && return 1
> fi
>
> if [ "$AC_UNKNOWN" == "CONTINUE" ]; then
> return 0 # assume on AC power
> elif [ "$AC_UNKNOWN" == "ABORT" ]; then
> return 1 # assume on battery
> else
> log err "Invalid value for AC_UNKNOWN in the config file"
> exit 1
> fi
> }
>
> More comments inline below.
>
> > diff --git a/scrub/e2scrub.in b/scrub/e2scrub.in
> > new file mode 100644
> > index 0000000..32d11c8
> > --- /dev/null
> > +++ b/scrub/e2scrub.in
> > @@ -0,0 +1,182 @@
> > +#!/bin/bash
> > +
> > +# Copyright (C) 2018 Oracle. All Rights Reserved.
> > +#
> > +# Author: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> > +#
> > +# This program is free software; you can redistribute it and/or
> > +# modify it under the terms of the GNU General Public License
> > +# as published by the Free Software Foundation; either version 2
> > +# of the License, or (at your option) any later version.
> > +#
> > +# This program is distributed in the hope that it would be useful,
> > +# but WITHOUT ANY WARRANTY; without even the implied warranty of
> > +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
> > +# GNU General Public License for more details.
> > +#
> > +# You should have received a copy of the GNU General Public License
> > +# along with this program; if not, write the Free Software Foundation,
> > +# Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA.
> > +
> > +# Automatically check a LVM-managed filesystem online.
> > +# We use lvm snapshots to do this, which means that we can only
> > +# check filesystems in VGs that have at least 256mb (or so) of
> > +# free space.
> > +
> > +snap_size_mb=256
> > +fstrim=0
> > +e2fsck_opts=""
> > +conffile="@root_sysconfdir@/e2scrub.conf"
> > +
> > +test -f "${conffile}" && . "${conffile}"
> > +
> > +print_help() {
> > + echo "Usage: $0 [-t] device"
> > + echo
> > + echo "device must be a LVM-managed block device"
> > + echo "-t: Run fstrim if successful."
> > +}
> > +
> > +exitcode() {
> > + ret="$1"
> > +
> > + exit "${ret}"
> > +}
> > +
> > +prog_path() {
> > + path="$1"
> > + displayname="$2"
> > +
> > + if ! type -P "${path}" && [ -n "${displayname}" ]; then
> > + echo "${displayname}: Command not found."
> > + exitcode 8
> > + fi
> > +}
> > +
> > +LVS_PROG="$(prog_path "@root_sbindir@/lvs" "lvs")"
> > +BLKID_PROG="$(prog_path "@root_sbindir@/blkid" "blkid")"
> > +LVCREATE_PROG="$(prog_path "@root_sbindir@/lvcreate" "lvcreate")"
> > +LVREMOVE_PROG="$(prog_path "@root_sbindir@/lvremove" "lvremove")"
> > +FSTRIM_PROG="$(prog_path "@root_sbindir@/fstrim")"
> > +UDEVADM_PROG="$(prog_path "@root_sbindir@/udevadm")"
> > +SLEEP_PROG="$(prog_path "@root_bindir@/sleep")"
>
> Rather than hard-coding the pathnames for all of these commands, it would be
> a lot easier to just use them by name and set an explicit PATH= to find them.
> That avoids the need to do all of the substitutions, and makes the script
> much less ugly.
Ok.
> > +while getopts "t" opt; do
> > + case "${opt}" in
> > + "t") fstrim=1;;
> > + *) print_help; exitcode 2;;
> > + esac
> > +done
> > +shift "$((OPTIND - 1))"
> > +
> > +dev="$1"
> > +if [ -z "${dev}" ]; then
> > + print_help
> > + exitcode 1
> > +elif [ ! -b "${dev}" ]; then
> > + echo "${dev}: Not a block device?"
> > + print_help
> > + exitcode 16
> > +fi
> > +
> > +# Make sure this is an LVM device we can snapshot
> > +vg="$("${LVS_PROG}" --noheadings -o vg_name "${dev}" 2> /dev/null | sed -e 's/^ //g')"
> > +lv="$("${LVS_PROG}" --noheadings -o lv_name "${dev}" 2> /dev/null | sed -e 's/^ //g')"
> > +if [ -z "${vg}" ] || [ -z "${lv}" ]; then
> > + echo "${dev}: Not a LVM device."
> > + exitcode 16
> > +fi
> > +start_time="$(date +'%Y%m%d%H%M%S')"
> > +snap="${lv}.e2scrub"
> > +snap_dev="/dev/${vg}/${snap}"
> > +fstype="$("${BLKID_PROG}" -p -s TYPE "${dev}" | sed -e 's/^.*TYPE="\(.*\)".*$/\1/g')"
> > +
> > +case "${fstype}" in
> > +"ext2"|"ext3"|"ext4")
> > + ;;
> > +*)
> > + echo "${dev}: Filesystem of type ${fstype} not supported."
> > + exitcode 16
> > + ;;
> > +esac
> > +
> > +teardown() {
> > + # Remove and wait for removal to succeed.
> > + "${LVREMOVE_PROG}" -f "${vg}/${snap}" 3>&-
> > + while [ -b "${snap_dev}" ] && [ "$?" -eq "5" ]; do
> > + /bin/sleep 0.5
> > + "${LVREMOVE_PROG}" -f "${vg}/${snap}" 3>&-
> > + done
> > +}
> > +
> > +check() {
> > + # First we preen the filesystem to recover the journal, then
> > + # we see if e2fsck tries any non-optimization repairs. If
> > + # either of these two returns a non-zero status (errors fixed
> > + # or remaining) then this fs is bad.
> > + E2FSCK_FIXES_ONLY=1
> > + export E2FSCK_FIXES_ONLY
> > + ${DBG} "@root_sbindir@/e2fsck" -p ${e2fsck_opts} "${snap_dev}" || return 1
> > + ${DBG} "@root_sbindir@/e2fsck" -fy ${e2fsck_opts} "${snap_dev}" || return
>
> This could run two full e2fsck checks for each filesystem. To recover only
> the journal, it would be better to use "e2fsck -E journal_only"?
>
> > +mark_corrupt() {
> > + ${DBG} "@root_sbindir@/tune2fs" -C 16000 -T "19000101" "${dev}"
>
> This won't actually do anything if the time/mount-based checks are disabled
> (which has been the default for a long time already, not that I agree with it).
> You need to add something like "-i 720" to force a check on the filesystem on
> the next mount.
The trouble with this (all of it, really) is that this clobbers whatever
setting the administrator might have written into the superblock. Given
that corruption failures will be logged and produce emailed reports, I
wonder if it would be easier to dispense with the force-fsck part
entirely?
--D
> Also, this clobbers the mount count unnecessarily.
>
> Finally, using "19000101" has the possibility to underflow in a different
> timezone, and is a negative signed timestamp as well (but is stored as unsigned
> value) so it may be problematic.
>
> Better to use "19700102" or similar, which is just as likely to cause a check.
>
> > +}
> > +
> > +setup() {
> > + # Create the snapshot, wait for device to appear
> > + teardown > /dev/null 2> /dev/null
> > + "${LVCREATE_PROG}" -s -L "${snap_size_mb}m" -n "${snap}" "${vg}/${lv}" 3>&-
> > + test $? -ne 0 && return 1
> > + test -x "${UDEVADM_PROG}" && "${UDEVADM_PROG}" settle
> > + return 0
> > +}
> > +
> > +trap "teardown" EXIT INT QUIT TERM
> > +if ! setup; then
> > + echo "Snapshot of ${dev} FAILED, will not check!"
> > + exitcode 1
> > +fi
> > +
> > +# Check and react
> > +if check; then
> > + echo "Scrub of ${dev} succeeded."
> > + mark_clean
> > +
> > + if [ "${fstrim}" -eq 1 ] && [ -x "${FSTRIM_PROG}" ]; then
> > + dir="$(lsblk -o MOUNTPOINT -n "${dev}")"
> > + if [ -d "${dir}" ]; then
> > + # NB: fstrim fails with snapshot present
> > + trap '' EXIT
> > + teardown
> > + "${FSTRIM_PROG}" -v "${dir}"
> > + fi
> > + fi
> > +
> > + ret=0
> > +else
> > + # fsck failed. Check if the snapshot is invalid; if so, make a
> > + # note of that at the end of the log. This isn't necessarily a
> > + # failure because the mounted fs could have overflowed the
> > + # snapshot with regular disk writes /or/ our repair process
> > + # could have done it by repairing too much.
> > + #
> > + # If it's really corrupt we ought to fsck at next boot.
> > + is_invalid="$("${LVS_PROG}" -o lv_snapshot_invalid --noheadings "${snap_dev}")"
> > + if [ -n "${is_invalid}" ]; then
> > + echo "Scrub of ${dev} FAILED due to invalid snapshot."
> > + ret=8
> > + else
> > + echo "Scrub of ${dev} FAILED! Reboot soon to fsck."
> > + mark_corrupt
> > + ret=6
> > + fi
> > +fi
> > +
> > +exitcode "${ret}"
>
>
> Cheers, Andreas
>
>
>
>
>
