starlingx/metal
Code Issues Proposed changes

Update crashDumpMgr to source config from envfile

Browse Source 
This commit updates the crashDumpMgr service in order to:
- Cleanup of current service naming and packaging to follow the
  standard Linux naming convention:
    - Repackage /etc/init.d/crashDumpMgr to
      /usr/sbin/crash-dump-manager
    - Rename crashDumpMgr.service to crash-dump-manager.service
- Add EnvironmentFile to crash-dump-manager service file to source
  configuration from /etc/default/crash-dump-manager.
- Update ExecStart of crash-dump-manager service to use parameters
  from EnvironmentFile
- Update crash-dump-manager service dependencies to run after
  config.service.
- Update logrotate configuration to support the retention polices of
  the maximum files. The “rotate 1” option was removed to permit
  crash-dump-manager to manage pruning old files.
- Modify the crash-dump-manager script to enable updates to the
  max_files parameter to a lower value. If there are currently more
  files than the new max_files value, the oldest files will be
  deleted the next time a crash dump file needs to be stored, thus
  adhering to the new max_files values.

Test Plan:

PASS: Build ISO and perform a fresh install. Verify the new
crash-dump-manager service is enabled and working as expected.
PASS: Add and apply new crashdump service parameters and force a kernel
panic. Verify that after the reboot, the max_files, max_used,
min_available and max_size values are updated accordingly to the service
parameters values.
PASS: Verify that the crashdump files are rotated as expected.

Story: 2010893
Task: 48910

Change-Id: I4a81fcc6ba456a0d73067b77588ee4a125e44e62
Signed-off-by: Enzo Candotti <enzo.candotti@windriver.com>
This commit is contained in:
Enzo Candotti 2023-10-06 15:45:30 -03:00
parent df8989e2a1
commit 23143abbca
11 changed files with 72 additions and 46 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
mtce/centos/mtce.spec
View File

@ -350,8 +350,8 @@ install -m 700 -p -D %{_buildsubdir}/alarm/scripts/mtcalarm.init %{buildroot}%{_
# TODO: Init hack. Should move to proper module
install -m 755 -p -D %{_buildsubdir}/scripts/hwclock.sh %{buildroot}%{_sysconfdir}/init.d/hwclock.sh
install -m 644 -p -D %{_buildsubdir}/scripts/hwclock.service %{buildroot}%{_unitdir}/hwclock.service
install -m 755 -p -D %{_buildsubdir}/scripts/crashDumpMgr %{buildroot}%{_sysconfdir}/init.d/crashDumpMgr
install -m 644 -p -D %{_buildsubdir}/scripts/crashDumpMgr.service %{buildroot}%{_unitdir}/crashDumpMgr.service
install -m 755 -p -D %{_buildsubdir}/scripts/crash-dump-manager %{buildroot}%{_sysconfdir}/init.d/crash-dump-manager
install -m 644 -p -D %{_buildsubdir}/scripts/crash-dump-manager.service %{buildroot}%{_unitdir}/crash-dump-manager.service
# systemd service files
install -m 644 -p -D %{_buildsubdir}/fsmon/scripts/fsmon.service %{buildroot}%{_unitdir}/fsmon.service
@ -435,7 +435,7 @@ install -m 755 -d %{buildroot}/var/run
/bin/systemctl enable rsyncd.service
/bin/systemctl enable goenabled.service
/bin/systemctl enable mtcalarm.service
/bin/systemctl enable crashDumpMgr.service
/bin/systemctl enable crash-dump-manager.service
%post -n mtce-hostw
/bin/systemctl enable hostw.service
@ -509,7 +509,7 @@ install -m 755 -d %{buildroot}/var/run
%{_sysconfdir}/init.d/mtcClient
%{_sysconfdir}/init.d/mtcalarm
%{_sysconfdir}/init.d/hwclock.sh
%{_sysconfdir}/init.d/crashDumpMgr
%{_sysconfdir}/init.d/crash-dump-manager
%{_unitdir}/runservices.service
%{_unitdir}/goenabled.service
@ -519,7 +519,7 @@ install -m 755 -d %{buildroot}/var/run
%{_unitdir}/mtcClient.service
%{_unitdir}/hbsClient.service
%{_unitdir}/hwclock.service
%{_unitdir}/crashDumpMgr.service
%{_unitdir}/crash-dump-manager.service
# Binaries
%{local_bindir}/mtcAgent

14
mtce/debian/deb_folder/crash-dump-manager.service Normal file
View File

@ -0,0 +1,14 @@
[Unit]
Description=Crash Dump Manager
After=network.target config.target
Before=sshd.service
DefaultEnabled=true
[Service]
Type=oneshot
RemainAfterExit=no
EnvironmentFile=/etc/default/crash-dump-manager
ExecStart=/usr/sbin/crash-dump-manager --max-size $MAX_SIZE --max-files $MAX_FILES --max-used $MAX_USED --min-available $MIN_AVAILABLE
[Install]
WantedBy=multi-user.target

12
mtce/debian/deb_folder/crashDumpMgr.service
View File

@ -1,12 +0,0 @@
[Unit]
Description=Crash Dump Manager
After=network.target
Before=sshd.service
[Service]
Type=oneshot
RemainAfterExit=no
ExecStart=/etc/init.d/crashDumpMgr --max-size 5Gi
[Install]
WantedBy=multi-user.target

4
mtce/debian/deb_folder/mtce.install
View File

@ -2,7 +2,7 @@ etc/bmc/server_profiles.d/sensor_hp360_v1_ilo_v4.profile
etc/bmc/server_profiles.d/sensor_hp380_v1_ilo_v4.profile
etc/bmc/server_profiles.d/sensor_quanta_v1_ilo_v4.profile
etc/collect.d/collect_bmc
etc/init.d/crashDumpMgr
etc/default/crash-dump-manager
etc/init.d/fsmon
etc/init.d/goenabled
etc/init.d/hbsClient
@ -22,6 +22,7 @@ etc/services.d/controller/mtcTest
etc/services.d/storage/mtcTest
etc/services.d/worker/mtcTest
etc/syslog-ng/conf.d/mtce.conf
etc/systemd/system-preset/
usr/lib/ocf/resource.d/platform/mtcAgent
usr/local/bin/fsmond
usr/local/bin/hbsAgent
@ -31,6 +32,7 @@ usr/local/bin/mtcClient
usr/local/bin/mtcalarmd
usr/local/bin/mtclogd
usr/local/bin/wipedisk
usr/sbin/crash-dump-manager
usr/sbin/dmemchk.sh
usr/sbin/fsync
usr/share/mtce/hwclock.service

9
mtce/debian/deb_folder/rules
View File

@ -36,6 +36,7 @@ override_dh_auto_install:
# Configuration files
install -m 600 -p -D scripts/mtc.ini $(SYSCONFDIR)/mtc.ini
install -m 600 -p -D scripts/crash-dump-manager_envfile $(SYSCONFDIR)/default/crash-dump-manager
install -m 600 -p -D scripts/mtc.conf $(SYSCONFDIR)/mtc.conf
install -m 600 -p -D fsmon/scripts/fsmond.conf $(SYSCONFDIR)/mtc/fsmond.conf
install -m 600 -p -D hwmon/scripts/hwmond.conf $(SYSCONFDIR)/mtc/hwmond.conf
@ -65,13 +66,13 @@ override_dh_auto_install:
install -m 755 -d $(SBINDIR)
install -m 700 -p -D fsync/fsync $(SBINDIR)/fsync
install -m 755 -d $(SBINDIR)
install -m 755 -p -D scripts/crash-dump-manager $(SBINDIR)/crash-dump-manager
install -m 700 -p -D pmon/scripts/pmon-restart $(LOCAL_SBINDIR)/pmon-restart
install -m 700 -p -D pmon/scripts/pmon-start $(LOCAL_SBINDIR)/pmon-start
install -m 700 -p -D pmon/scripts/pmon-stop $(LOCAL_SBINDIR)/pmon-stop
# init script files
install -m 755 -d $(INITDIR)
install -m 755 -p -D scripts/crashDumpMgr $(INITDIR)/crashDumpMgr
install -m 700 -p -D scripts/mtcClient $(INITDIR)/mtcClient
install -m 700 -p -D scripts/hbsClient $(INITDIR)/hbsClient
install -m 700 -p -D hwmon/scripts/lsb/hwmon $(INITDIR)/hwmon
@ -148,8 +149,12 @@ override_dh_auto_install:
install -m 755 -d $(ROOT)/var
install -m 755 -d $(ROOT)/var/run
# Presets
install -d $(SYSCONFDIR)/systemd/system-preset
install -m 544 scripts/55-crash-dump-manager.preset $(SYSCONFDIR)/systemd/system-preset
override_dh_installsystemd:
dh_installsystemd --name crashDumpMgr
dh_installsystemd --name crash-dump-manager
dh_installsystemd --name fsmon
dh_installsystemd --name goenabled
dh_installsystemd --name hbsClient

1
mtce/src/scripts/55-crash-dump-manager.preset Normal file
View File

@ -0,0 +1 @@
enable crash-dump-manager.service

33
mtce/src/scripts/crashDumpMgr → mtce/src/scripts/crash-dump-manager
View File

@ -5,7 +5,7 @@
# SPDX-License-Identifier: Apache-2.0
#
CRASHDUMPMGR_TAG=${CRASHDUMPMGR_TAG:-"crashDumpMgr"}
CRASHDUMPMGR_TAG=${CRASHDUMPMGR_TAG:-"crash-dump-manager"}
RETVAL=0
@ -13,6 +13,7 @@ RETVAL=0
DEFAULT_MAX_SIZE=5368709120 # "5GiB"
DEFAULT_MAX_FILES=4
UNLIMITED="unlimited"
DEFAULT="default"
DEFAULT_MAX_USED="${UNLIMITED}" # Assign UNLIMITED to DEFAULT_MAX_USED
DEFAULT_MIN_REMAINDER_PERCENT=10
MIN_REMAINDER_MINIMUM=1073741824 # 1GiB in bytes
@ -141,9 +142,15 @@ function manage_crash_dumps()
vmcore_size_hr=$(${NUMFMT_TO_HR} ${vmcore_size})
# Manage max number of files
if [ "$(ls -A ${CRASH_BUNDLE_DIR} | wc -l)" -ge "${max_files}" ]; then
oldest_vmcore=$(ls -t ${CRASH_BUNDLE_DIR} | tail -2 | head -1)
oldest_vmcore_size=$(stat --format='%s' ${oldest_vmcore})
files_in_crash_bundle_dir=$(ls -A ${CRASH_BUNDLE_DIR} | wc -l)
num_files_to_remove=$((files_in_crash_bundle_dir-max_files+1))
if [ "${num_files_to_remove}" -ge 1 ]; then
files_to_remove=$(ls -t ${CRASH_BUNDLE_DIR} | tail -$((num_files_to_remove+1)) | head -$((num_files_to_remove)))
files_to_remove_size=0
for file in ${files_to_remove}; do
files_to_remove_size+=$(stat --format='%s' ${file})
done
max_files_saved=true
fi
@ -156,8 +163,8 @@ function manage_crash_dumps()
# if the ${CRASH_BUNDLE_DIR} contains the maximum number of files, the available and used_space
# need to be updated to the value after deleting the oldest crash dump file.
if [ "${max_files_saved}" = true ] ; then
available=$((available+oldest_vmcore_size))
used_space=$((used_space-oldest_vmcore_size))
available=$((available+files_to_remove_size))
used_space=$((used_space-files_to_remove_size))
fi
available_hr=$(${NUMFMT_TO_HR} ${available})
log "new vmcore detected (size:${vmcore_size}:${vmcore_size_hr}) ;" \
@ -194,9 +201,11 @@ function manage_crash_dumps()
else
if [ "${max_files_saved}" = true ] ; then
# delete oldest vmcore file
log "removing oldest vmcore file: ${oldest_vmcore}"
rm -rf "${CRASH_BUNDLE_DIR}/${oldest_vmcore}"
for file in ${files_to_remove}; do
# delete old vmcore file
log "removing old vmcore file: ${file}"
rm -rf "${CRASH_BUNDLE_DIR}/${file}"
done
fi
log "creating bundle from ${entry}"
${IONICE_CMD} ${NICE_CMD} ${TAR_CMD} ${OTHER_BUNDLE}_${time}.tar -C ${CRASH_DIR} $(basename ${entry})
@ -284,7 +293,11 @@ while [[ $# -gt 0 ]]; do
--min-available)
shift
min_available=$(${NUMFMT_FROM_HR} "$(normalize_size_format "${1}")")
if [ "${1}" = "${DEFAULT}" ]; then
min_available=""
else
min_available=$(${NUMFMT_FROM_HR} "$(normalize_size_format "${1}")")
fi
shift
;;

13
mtce/src/scripts/crash-dump-manager.service Normal file
View File

@ -0,0 +1,13 @@
[Unit]
Description=Crash Dump Manager
After=network.target config.target
Before=sshd.service
[Service]
Type=oneshot
RemainAfterExit=no
EnvironmentFile=/etc/default/crash-dump-manager
ExecStart=/usr/sbin/crash-dump-manager --max-size $MAX_SIZE --max-files $MAX_FILES --max-used $MAX_USED --min-available $MIN_AVAILABLE
[Install]
WantedBy=multi-user.target

4
mtce/src/scripts/crash-dump-manager_envfile Normal file
View File

@ -0,0 +1,4 @@
MAX_FILES=4
MAX_SIZE=5Gi
MAX_USED=unlimited
MIN_AVAILABLE=default

12
mtce/src/scripts/crashDumpMgr.service
View File

@ -1,12 +0,0 @@
[Unit]
Description=Crash Dump Manager
After=network.target
Before=sshd.service
[Service]
Type=oneshot
RemainAfterExit=no
ExecStart=/etc/init.d/crashDumpMgr --max-size 5Gi
[Install]
WantedBy=multi-user.target

6
mtce/src/scripts/crashdump.logrotate
View File

@ -1,14 +1,12 @@
#
# Copyright (c) 2020-2021 Wind River Systems, Inc.
# Copyright (c) 2020-2021,2023 Wind River Systems, Inc.
#
# SPDX-License-Identifier: Apache-2.0
/var/log/crash/vmcore.tar
/var/log/crash/vmcore_first.tar
/var/log/crash/vmcore_*.tar
{
size 1K
start 1
rotate 1
missingok
notifempty
compress