diff options
Diffstat (limited to 'extraconfig/tasks')
-rwxr-xr-x | extraconfig/tasks/major_upgrade_ceph_mon.sh | 8 | ||||
-rw-r--r-- | extraconfig/tasks/major_upgrade_ceph_storage.sh | 2 | ||||
-rwxr-xr-x | extraconfig/tasks/major_upgrade_check.sh | 104 | ||||
-rwxr-xr-x | extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh | 68 | ||||
-rw-r--r-- | extraconfig/tasks/major_upgrade_pacemaker.yaml | 22 | ||||
-rw-r--r-- | extraconfig/tasks/major_upgrade_pacemaker_init.j2.yaml (renamed from extraconfig/tasks/major_upgrade_pacemaker_init.yaml) | 78 | ||||
-rw-r--r-- | extraconfig/tasks/mitaka_to_newton_aodh_data_migration.yaml | 12 |
7 files changed, 142 insertions, 152 deletions
diff --git a/extraconfig/tasks/major_upgrade_ceph_mon.sh b/extraconfig/tasks/major_upgrade_ceph_mon.sh index b76dd7c3..21a2b5bc 100755 --- a/extraconfig/tasks/major_upgrade_ceph_mon.sh +++ b/extraconfig/tasks/major_upgrade_ceph_mon.sh @@ -18,13 +18,13 @@ if ! [[ "$INSTALLED_VERSION" =~ ^0\.94.* ]]; then fi CEPH_STATUS=$(ceph health | awk '{print $1}') -if [ ${CEPH_STATUS} = HEALTH_ERR ]; do +if [ ${CEPH_STATUS} = HEALTH_ERR ]; then echo ERROR: Ceph cluster status is HEALTH_ERR, cannot be upgraded exit 1 fi # Useful when upgrading with OSDs num < replica size -if [ $ignore_ceph_upgrade_warnings != "true" ]; then +if [ ${ignore_ceph_upgrade_warnings:-false} != "true" ]; then timeout 300 bash -c "while [ ${CEPH_STATUS} != HEALTH_OK ]; do echo WARNING: Waiting for Ceph cluster status to go HEALTH_OK; sleep 30; @@ -44,7 +44,7 @@ timeout 60 bash -c "while kill -0 ${MON_PID} 2> /dev/null; do done" # Update to Jewel -yum -y -q update ceph-mon +yum -y -q update ceph-mon ceph # Restart/Exit if not on Jewel, only in that case we need the changes UPDATED_VERSION=$(ceph --version | awk '{print $3}') @@ -54,7 +54,7 @@ if [[ "$UPDATED_VERSION" =~ ^0\.94.* ]]; then elif [[ "$UPDATED_VERSION" =~ ^10\.2.* ]]; then # RPM could own some of these but we can't take risks on the pre-existing files for d in /var/lib/ceph/mon /var/log/ceph /var/run/ceph /etc/ceph; do - chown -R ceph:ceph $d + chown -R ceph:ceph $d || echo WARNING: chown of $d failed done # Replay udev events with newer rules diff --git a/extraconfig/tasks/major_upgrade_ceph_storage.sh b/extraconfig/tasks/major_upgrade_ceph_storage.sh index 03a1c1c2..dc80a724 100644 --- a/extraconfig/tasks/major_upgrade_ceph_storage.sh +++ b/extraconfig/tasks/major_upgrade_ceph_storage.sh @@ -63,7 +63,7 @@ if [[ "$UPDATED_VERSION" =~ ^0\.94.* ]]; then elif [[ "$UPDATED_VERSION" =~ ^10\.2.* ]]; then # RPM could own some of these but we can't take risks on the pre-existing files for d in /var/lib/ceph/osd /var/log/ceph /var/run/ceph /etc/ceph; do - chown -R ceph:ceph $d + chown -R ceph:ceph $d || echo WARNING: chown of $d failed done # Replay udev events with newer rules diff --git a/extraconfig/tasks/major_upgrade_check.sh b/extraconfig/tasks/major_upgrade_check.sh new file mode 100755 index 00000000..dc7ec71a --- /dev/null +++ b/extraconfig/tasks/major_upgrade_check.sh @@ -0,0 +1,104 @@ +#!/bin/bash + +set -eu + +check_cluster() +{ + if pcs status 2>&1 | grep -E '(cluster is not currently running)|(OFFLINE:)'; then + echo_error "ERROR: upgrade cannot start with some cluster nodes being offline" + exit 1 + fi +} + +check_pcsd() +{ + if pcs status 2>&1 | grep -E 'Offline'; then + echo_error "ERROR: upgrade cannot start with some pcsd daemon offline" + exit 1 + fi +} + +check_disk_for_mysql_dump() +{ + # Where to backup current database if mysql need to be upgraded + MYSQL_BACKUP_DIR=/var/tmp/mysql_upgrade_osp + MYSQL_TEMP_UPGRADE_BACKUP_DIR=/var/lib/mysql-temp-upgrade-backup + # Spare disk ratio for extra safety + MYSQL_BACKUP_SIZE_RATIO=1.2 + + # Shall we upgrade mysql data directory during the stack upgrade? + if [ "$mariadb_do_major_upgrade" = "auto" ]; then + ret=$(is_mysql_upgrade_needed) + if [ $ret = "1" ]; then + DO_MYSQL_UPGRADE=1 + else + DO_MYSQL_UPGRADE=0 + fi + echo "mysql upgrade required: $DO_MYSQL_UPGRADE" + elif [ "$mariadb_do_major_upgrade" = "no" ]; then + DO_MYSQL_UPGRADE=0 + else + DO_MYSQL_UPGRADE=1 + fi + + if [ "$(hiera -c /etc/puppet/hiera.yaml bootstrap_nodeid)" = "$(facter hostname)" ]; then + if [ $DO_MYSQL_UPGRADE -eq 1 ]; then + + if [ -d "$MYSQL_BACKUP_DIR" ]; then + echo_error "Error: $MYSQL_BACKUP_DIR exists already. Likely an upgrade failed previously" + exit 1 + fi + mkdir "$MYSQL_BACKUP_DIR" + if [ $? -ne 0 ]; then + echo_error "Error: could not create temporary backup directory $MYSQL_BACKUP_DIR" + exit 1 + fi + + # the /root/.my.cnf is needed because we set the mysql root + # password from liberty onwards + backup_flags="--defaults-extra-file=/root/.my.cnf -u root --flush-privileges --all-databases --single-transaction" + # While not ideal, this step allows us to calculate exactly how much space the dump + # will need. Our main goal here is avoiding any chance of corruption due to disk space + # exhaustion + backup_size=$(mysqldump $backup_flags 2>/dev/null | wc -c) + database_size=$(du -cb /var/lib/mysql | tail -1 | awk '{ print $1 }') + free_space=$(df -B1 --output=avail "$MYSQL_BACKUP_DIR" | tail -1) + + # we need at least space for a new mysql database + dump of the existing one, + # times a small factor for additional safety room + # note: bash doesn't do floating point math or floats in if statements, + # so use python to apply the ratio and cast it back to integer + required_space=$(python -c "from __future__ import print_function; print(\"%d\" % int((($database_size + $backup_size) * $MYSQL_BACKUP_SIZE_RATIO)))") + if [ $required_space -ge $free_space ]; then + echo_error "Error: not enough free space in $MYSQL_BACKUP_DIR ($required_space bytes required)" + exit 1 + fi + fi + fi +} + +check_python_rpm() +{ + # If for some reason rpm-python are missing we want to error out early enough + if ! rpm -q rpm-python &> /dev/null; then + echo_error "ERROR: upgrade cannot start without rpm-python installed" + exit 1 + fi +} + +check_clean_cluster() +{ + if crm_mon -1 | grep -A3 Failed; then + echo_error "ERROR: upgrade cannot start with failed resources on the cluster. Clean them up before starting: pcs resource cleanup." + exit 1 + fi +} + +check_galera_root_password() +{ + # BZ: 1357112 + if [ ! -e /root/.my.cnf ]; then + echo_error "ERROR: upgrade cannot be started, the galera password is missing. The overcloud needs update." + exit 1 + fi +} diff --git a/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh b/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh index 0b702630..e81ca086 100755 --- a/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh +++ b/extraconfig/tasks/major_upgrade_controller_pacemaker_1.sh @@ -4,11 +4,12 @@ set -eu cluster_sync_timeout=1800 -if pcs status 2>&1 | grep -E '(cluster is not currently running)|(OFFLINE:)'; then - echo_error "ERROR: upgrade cannot start with some cluster nodes being offline" - exit 1 -fi - +check_cluster +check_pcsd +check_clean_cluster +check_python_rpm +check_galera_root_password +check_disk_for_mysql_dump # We want to disable fencing during the cluster --stop as it might fence # nodes where a service fails to stop, which could be fatal during an upgrade @@ -17,12 +18,6 @@ fi STONITH_STATE=$(pcs property show stonith-enabled | grep "stonith-enabled" | awk '{ print $2 }') pcs property set stonith-enabled=false -# If for some reason rpm-python are missing we want to error out early enough -if ! rpm -q rpm-python &> /dev/null; then - echo_error "ERROR: upgrade cannot start without rpm-python installed" - exit 1 -fi - # In case the mysql package is updated, the database on disk must be # upgraded as well. This typically needs to happen during major # version upgrades (e.g. 5.5 -> 5.6, 5.5 -> 10.1...) @@ -35,59 +30,8 @@ fi # on mysql package versionning, but this can be overriden manually # to support specific upgrade scenario -# Where to backup current database if mysql need to be upgraded -MYSQL_BACKUP_DIR=/var/tmp/mysql_upgrade_osp -MYSQL_TEMP_UPGRADE_BACKUP_DIR=/var/lib/mysql-temp-upgrade-backup -# Spare disk ratio for extra safety -MYSQL_BACKUP_SIZE_RATIO=1.2 - -# Shall we upgrade mysql data directory during the stack upgrade? -if [ "$mariadb_do_major_upgrade" = "auto" ]; then - ret=$(is_mysql_upgrade_needed) - if [ $ret = "1" ]; then - DO_MYSQL_UPGRADE=1 - else - DO_MYSQL_UPGRADE=0 - fi - echo "mysql upgrade required: $DO_MYSQL_UPGRADE" -elif [ "$mariadb_do_major_upgrade" = "no" ]; then - DO_MYSQL_UPGRADE=0 -else - DO_MYSQL_UPGRADE=1 -fi - if [ "$(hiera -c /etc/puppet/hiera.yaml bootstrap_nodeid)" = "$(facter hostname)" ]; then if [ $DO_MYSQL_UPGRADE -eq 1 ]; then - if [ -d "$MYSQL_BACKUP_DIR" ]; then - echo_error "Error: $MYSQL_BACKUP_DIR exists already. Likely an upgrade failed previously" - exit 1 - fi - mkdir "$MYSQL_BACKUP_DIR" - if [ $? -ne 0 ]; then - echo_error "Error: could not create temporary backup directory $MYSQL_BACKUP_DIR" - exit 1 - fi - - # the /root/.my.cnf is needed because we set the mysql root - # password from liberty onwards - backup_flags="--defaults-extra-file=/root/.my.cnf -u root --flush-privileges --all-databases --single-transaction" - # While not ideal, this step allows us to calculate exactly how much space the dump - # will need. Our main goal here is avoiding any chance of corruption due to disk space - # exhaustion - backup_size=$(mysqldump $backup_flags 2>/dev/null | wc -c) - database_size=$(du -cb /var/lib/mysql | tail -1 | awk '{ print $1 }') - free_space=$(df -B1 --output=avail "$MYSQL_BACKUP_DIR" | tail -1) - - # we need at least space for a new mysql database + dump of the existing one, - # times a small factor for additional safety room - # note: bash doesn't do floating point math or floats in if statements, - # so use python to apply the ratio and cast it back to integer - required_space=$(python -c "from __future__ import print_function; print(\"%d\" % int((($database_size + $backup_size) * $MYSQL_BACKUP_SIZE_RATIO)))") - if [ $required_space -ge $free_space ]; then - echo_error "Error: not enough free space in $MYSQL_BACKUP_DIR ($required_space bytes required)" - exit 1 - fi - mysqldump $backup_flags > "$MYSQL_BACKUP_DIR/openstack_database.sql" cp -rdp /etc/my.cnf* "$MYSQL_BACKUP_DIR" fi diff --git a/extraconfig/tasks/major_upgrade_pacemaker.yaml b/extraconfig/tasks/major_upgrade_pacemaker.yaml index 598d22d0..7244f949 100644 --- a/extraconfig/tasks/major_upgrade_pacemaker.yaml +++ b/extraconfig/tasks/major_upgrade_pacemaker.yaml @@ -1,16 +1,8 @@ -heat_template_version: 2014-10-16 +heat_template_version: 2016-10-14 description: 'Upgrade for Pacemaker deployments' parameters: - controller_servers: - type: json - compute_servers: - type: json - blockstorage_servers: - type: json - objectstorage_servers: - type: json - cephstorage_servers: + servers: type: json input_values: type: json @@ -54,9 +46,10 @@ resources: CephMonUpgradeDeployment: type: OS::Heat::SoftwareDeploymentGroup properties: - servers: {get_param: controller_servers} + servers: {get_param: servers, Controller} config: {get_resource: CephMonUpgradeConfig} input_values: {get_param: input_values} + update_policy: batch_create: max_batch_size: 1 rolling_update: @@ -82,6 +75,7 @@ resources: params: MYSQL_MAJOR_UPGRADE: {get_param: MySqlMajorUpgrade} - get_file: pacemaker_common_functions.sh + - get_file: major_upgrade_check.sh - get_file: major_upgrade_pacemaker_migrations.sh - get_file: major_upgrade_controller_pacemaker_1.sh @@ -89,7 +83,7 @@ resources: type: OS::Heat::SoftwareDeploymentGroup depends_on: CephMonUpgradeDeployment properties: - servers: {get_param: controller_servers} + servers: {get_param: servers, Controller} config: {get_resource: ControllerPacemakerUpgradeConfig_Step1} input_values: {get_param: input_values} @@ -103,7 +97,7 @@ resources: BlockStorageUpgradeDeployment: type: OS::Heat::SoftwareDeploymentGroup properties: - servers: {get_param: blockstorage_servers} + servers: {get_param: servers, BlockStorage} config: {get_resource: BlockStorageUpgradeConfig} input_values: {get_param: input_values} @@ -122,7 +116,7 @@ resources: type: OS::Heat::SoftwareDeploymentGroup depends_on: BlockStorageUpgradeDeployment properties: - servers: {get_param: controller_servers} + servers: {get_param: servers, Controller} config: {get_resource: ControllerPacemakerUpgradeConfig_Step2} input_values: {get_param: input_values} diff --git a/extraconfig/tasks/major_upgrade_pacemaker_init.yaml b/extraconfig/tasks/major_upgrade_pacemaker_init.j2.yaml index 623549a0..f6aa3066 100644 --- a/extraconfig/tasks/major_upgrade_pacemaker_init.yaml +++ b/extraconfig/tasks/major_upgrade_pacemaker_init.j2.yaml @@ -3,15 +3,7 @@ description: 'Upgrade for Pacemaker deployments' parameters: - controller_servers: - type: json - compute_servers: - type: json - blockstorage_servers: - type: json - objectstorage_servers: - type: json - cephstorage_servers: + servers: type: json input_values: type: json @@ -43,45 +35,12 @@ resources: - "if [[ -f /etc/resolv.conf.save ]] ; then rm /etc/resolv.conf.save; fi\n\n" - get_param: UpgradeInitCommand - UpgradeInitControllerDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: controller_servers} - config: {get_resource: UpgradeInitConfig} - input_values: {get_param: input_values} - - UpgradeInitComputeDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: compute_servers} - config: {get_resource: UpgradeInitConfig} - input_values: {get_param: input_values} - - UpgradeInitBlockStorageDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: blockstorage_servers} - config: {get_resource: UpgradeInitConfig} - input_values: {get_param: input_values} - - UpgradeInitObjectStorageDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: objectstorage_servers} - config: {get_resource: UpgradeInitConfig} - input_values: {get_param: input_values} - - UpgradeInitCephStorageDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: cephstorage_servers} - config: {get_resource: UpgradeInitConfig} - input_values: {get_param: input_values} - # TODO(jistr): for Mitaka->Newton upgrades and further we can use # map_merge with input_values instead of feeding params into scripts # via str_replace on bash snippets + # FIXME(shardy) we have hard-coded per-role *ScriptConfig's here + # Would be better to have a common config for all roles ComputeDeliverUpgradeScriptConfig: type: OS::Heat::SoftwareConfig properties: @@ -97,35 +56,32 @@ resources: UPGRADE_LEVEL_NOVA_COMPUTE: {get_param: UpgradeLevelNovaCompute} - get_file: major_upgrade_compute.sh - ComputeDeliverUpgradeScriptDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: compute_servers} - config: {get_resource: ComputeDeliverUpgradeScriptConfig} - input_values: {get_param: input_values} - ObjectStorageDeliverUpgradeScriptConfig: type: OS::Heat::SoftwareConfig properties: group: script config: {get_file: major_upgrade_object_storage.sh} - ObjectStorageDeliverUpgradeScriptDeployment: - type: OS::Heat::SoftwareDeploymentGroup - properties: - servers: {get_param: objectstorage_servers} - config: {get_resource: ObjectStorageDeliverUpgradeScriptConfig} - input_values: {get_param: input_values} - CephStorageDeliverUpgradeScriptConfig: type: OS::Heat::SoftwareConfig properties: group: script config: {get_file: major_upgrade_ceph_storage.sh} - CephStorageDeliverUpgradeScriptDeployment: +{% for role in roles %} + UpgradeInit{{role.name}}Deployment: + type: OS::Heat::SoftwareDeploymentGroup + properties: + servers: {get_param: [servers, {{role.name}}]} + config: {get_resource: UpgradeInitConfig} + input_values: {get_param: input_values} + + {% if not role.name in ['Controller', 'BlockStorage'] %} + {{role.name}}DeliverUpgradeScriptDeployment: type: OS::Heat::SoftwareDeploymentGroup properties: - servers: {get_param: cephstorage_servers} - config: {get_resource: CephStorageDeliverUpgradeScriptConfig} + servers: {get_param: [servers, {{role.name}}]} + config: {get_resource: {{role.name}}DeliverUpgradeScriptConfig} input_values: {get_param: input_values} + {% endif %} +{% endfor %} diff --git a/extraconfig/tasks/mitaka_to_newton_aodh_data_migration.yaml b/extraconfig/tasks/mitaka_to_newton_aodh_data_migration.yaml index 9414ac19..91406fba 100644 --- a/extraconfig/tasks/mitaka_to_newton_aodh_data_migration.yaml +++ b/extraconfig/tasks/mitaka_to_newton_aodh_data_migration.yaml @@ -4,15 +4,7 @@ description: > Software-config for performing aodh data migration parameters: - controller_servers: - type: json - compute_servers: - type: json - blockstorage_servers: - type: json - objectstorage_servers: - type: json - cephstorage_servers: + servers: type: json input_values: type: json @@ -28,6 +20,6 @@ resources: AodhMysqlMigrationScriptDeployment: type: OS::Heat::SoftwareDeploymentGroup properties: - servers: {get_param: controller_servers} + servers: {get_param: servers, Controller} config: {get_resource: AodhMysqlMigrationScriptConfig} input_values: {get_param: input_values} |