Debian: KS: Add worker system node install support to kickstart

This update adds kickstart support for the following install types
 - worker
 - worker lowlatency

This update also
 - Adds install type prefix to all kickstart logs.
 - Fixes an issue that was causing all the kickstart logs produced
   with 'ilog' to have the same date.
 - Fixes the report_failure_with_msg handling.
 - Removed defaulting to aio if traits boot argument is missing.
 - Hardens handling of invalid traits and invalid trait combinations.
   Install will fail if traits arg is missing or if an invalid
   trait combination is specified.

Test Plan:

PASS: Verify Debian Build packages and image
PASS: Verify worker node install and disk boot
PASS: Verify lowlatency worker node install and disk boot
PASS: Verify install type prefix added to logging
PASS: Verify kernel arguements
PASS: Verify no platform backup partition is created on
      worker only node
PASS: Verify invalid traits handling

Regression:

PASS: Verify AIO standard install and enable (c0 & c1)
PASS: Verify AIO lowlatency install and enable (c0 & c1)
PASS: Verify standard controller install and enable
PASS: Verify platform-backup partition create and persistence
PASS: Verify system host lock/unlock controller-1
PASS; Verify system host-reinstall controller-1
PASS: Verify logging

Story: 2009968
Task: 45831
Signed-off-by: Eric MacDonald <eric.macdonald@windriver.com>
Change-Id: I6446e5b94b2b386aa0ac1d92e5ffb2cb1167db27
This commit is contained in:
Eric MacDonald 2022-07-26 15:29:03 +00:00
parent af907fd960
commit 004eea6acd

View File

@ -20,7 +20,6 @@
# storage ... the storage function
# worker ... the worker function
# lowlatency ... the low latency function
# miniboot ... redfish subcloud install function
#
# Valid node personality trait groupings ... set configuration function
#
@ -98,24 +97,24 @@ export LOGFILE="/\${LAT_DIR}/kickstart.log"
function log()
{
local dt="\$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} info: \$1" >>\${LOGFILE}
echo "\$dt kickstart \${HOOK_LABEL} info: \${prefix} \$1" >>\${LOGFILE}
}
function ilog()
{
[ -z "$stdout" ] && stdout=1
local dt="$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} info: \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} info: \$1" >>\${LOGFILE}
[ -z "\${stdout}" ] && stdout=1
local dt="\$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} info: \${prefix} \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} info: \${prefix} \$1" >>\${LOGFILE}
}
function dlog()
{
if [ "\${debug}" ] ; then
if [ \${debug} -ne 0 ] ; then
[ -z "\$stdout" ] && stdout=1
local dt="\$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} debug: \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} debug: \$1" >>\${LOGFILE}
echo "\$dt kickstart \${HOOK_LABEL} debug: \${prefix} \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} debug: \${prefix} \$1" >>\${LOGFILE}
fi
}
@ -123,24 +122,34 @@ function wlog()
{
[ -z "\$stdout" ] && stdout=1
local dt="\$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} warn: \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} warn: \$1" >>\${LOGFILE}
echo "\$dt kickstart \${HOOK_LABEL} warn: \${prefix} \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} warn: \${prefix} \$1" >>\${LOGFILE}
}
function elog()
{
[ -z "\$stdout" ] && stdout=1
local dt="\$(date "+%Y-%m-%d %H:%M:%S.%3N")"
echo "\$dt kickstart \${HOOK_LABEL} error: \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} error: \$1" >>\${LOGFILE}
echo "\$dt kickstart \${HOOK_LABEL} error: \${prefix} \$1" >&\${stdout}
echo "\$dt kickstart \${HOOK_LABEL} error: \${prefix} \$1" >>\${LOGFILE}
}
########################################################################
# Name : report_failure_with_msg
# Purpose : Report installation error, offer a console and
# reboot on exit.
# Parameters: \$1 is the failure message string
# Return : Does not return
########################################################################
function report_failure_with_msg()
{
local msg=\${1}
echo -e '\n\nInstallation failed.\n'
wlog "Installation Failed: ERROR: \${msg}"
exit 1
echo "... dropping to shell ; exit to reboot ..."
sleep 1
bash
reboot -f
}
########################################################################
@ -219,12 +228,12 @@ function exec_retry()
local ret_code=0
local ret_stdout=""
command="\${command}" # 2>&\$stdout"
while [ \$retry_count -le \$retries ]; do
command="\${command}" # 2>&\${stdout}"
while [ \${retry_count} -le \${retries} ]; do
# dlog "Running command: '\${command}'."
ret_stdout=\$(eval \${command})
ret_code=\${?}
[ \$ret_code -eq 0 ] && break
[ \${ret_code} -eq 0 ] && break
wlog "Error running command '\${command}'. Try \${retry_count} of \${retries} retry in \${interval}s."
wlog "ret_code: \${ret_code}, stdout: '\${ret_stdout}'."
sleep \${interval}
@ -254,16 +263,16 @@ function exec_no_fds()
local ret_code=0
local ret_stdout=""
for fd in \$fds
for fd in \${fds}
do
local cmd="\$cmd \$fd>&-"
local cmd="\${cmd} \$fd>&-"
done
if [ -z "\$retries" ]; then
#wlog "Running command: '\$cmd'."
eval "\$cmd"
if [ -z "\${retries}" ]; then
#wlog "Running command: '\${cmd}'."
eval "\${cmd}"
else
ret_stdout=\$(exec_retry "\$retries" "\$interval" "\$cmd")
ret_code=\$?
ret_stdout=\$(exec_retry "\${retries}" "\${interval}" "\${cmd}")
ret_code=\${?}
echo "\${ret_stdout}"
return \${ret_code}
fi
@ -271,11 +280,27 @@ function exec_no_fds()
#########################################################################
# Name : is_system_node_install
# Returns : 1 if yes 'pxecontroller' is in $insturl boot arg ; else 0
# Returns : true : 0 if 'pxecontroller' is in $insturl boot arg
# false: 1
#########################################################################
function is_system_node_install()
{
if [[ "\$insturl" == *"pxecontroller"* ]] ; then
if [[ "\${insturl}" == *"pxecontroller"* ]] ; then
return 0
else
return 1
fi
}
#########################################################################
# Name : is_usb_install
# Returns : true :0
# false:1
#########################################################################
function is_usb_install()
{
pxeurl=\$(echo \${insturl} | sed -e s/ostree_repo//)
if [ "\${pxeurl}" = "file://NOT_SET" ] ; then
return 0
else
return 1
@ -342,19 +367,19 @@ function get_iface_from_ethname()
function display_volume_info ()
{
if [ \${debug} -ne 0 ] ; then
ilog "Volume Info:\${1}"
ilog "Volume Info:\${1}"
pvscan 2>/dev/null
pvscan 2>/dev/null >> \${LOGFILE}
pvscan 2>/dev/null
pvscan 2>/dev/null >> \${LOGFILE}
lvscan 2>/dev/null
lvscan 2>/dev/null >> \${LOGFILE}
lvscan 2>/dev/null
lvscan 2>/dev/null >> \${LOGFILE}
pvdisplay 2>/dev/null
pvdisplay 2>/dev/null >> \${LOGFILE}
pvdisplay 2>/dev/null
pvdisplay 2>/dev/null >> \${LOGFILE}
lvdisplay 2>/dev/null
lvdisplay 2>/dev/null >> \${LOGFILE}
lvdisplay 2>/dev/null
lvdisplay 2>/dev/null >> \${LOGFILE}
fi
}
@ -406,7 +431,7 @@ function clr_variable()
#########################################################################
function get_aio_controller_provisioning_sizes()
{
gb=$((1024*1024*1024))
gb=\$((1024*1024*1024))
##
## NOTE: updates to partition sizes need to be also reflected in
@ -506,11 +531,11 @@ function get_aio_controller_provisioning_sizes()
ilog "install disk size:\$sz gb:\$gb"
# Round MINIMUM_PLATFORM_PV_SIZE to the closest upper value that can be divided by 1024.
if [ \$sz -gt \$((\$default_small_disk_size*\$gb)) ] ; then
if [ \${sz} -gt \$((\${default_small_disk_size}*\${gb})) ] ; then
ilog "Large disk: MINIMUM_PLATFORM_PV_SIZE=179GiB*1024=183296"
export MINIMUM_PLATFORM_PV_SIZE=183296
elif [ \$sz -ge \$((\$minimum_small_disk_size*\$gb)) ] ; then
elif [ \${sz} -ge \$((\${minimum_small_disk_size}*\${gb})) ] ; then
ilog "Small disk: MINIMUM_PLATFORM_PV_SIZE=164GiB*1024=167936"
export MINIMUM_PLATFORM_PV_SIZE=167936
@ -536,6 +561,30 @@ function get_std_controller_provisioning_sizes()
export SCRATCH_VOL_SIZE=16000
}
function get_worker_provisioning_sizes()
{
gb=\$((1024*1024*1024))
export LOG_VOL_SIZE=4000
export SCRATCH_VOL_SIZE=4000
## LOG_VOL_SIZE = 4096
## SCRATCH_VOL_SIZE = 4096
## DOCKER = 30720
## CEPH_MON = 20480
## KUBELET_VOL_SIZE = 10240
## RESERVED_PE = 16 (based on pesize=32768)
##
## CGTS_PV_SIZE = 4096 + 4096 + 30720 + 20480 + 10240 + 16 = 69648
##
## Round CGTS_PV_SIZE to the closest upper value that can be divided by 1024.
## 69648/1024=68.01. CGTS_PV_SIZE=69*1024=70656.
export CGTS_PV_SIZE=70656
sz=\$(blockdev --getsize64 \${INSTDEV})
ilog "install disk size:\${sz} gb:\${gb}"
}
##########################################################################
# Global Kickstart Constants #
@ -544,28 +593,30 @@ function get_std_controller_provisioning_sizes()
# NOTE: exported Variables from other segments are NOT visible from functions
#
if [ "${INSTFLUX}" = 0 ] ; then
BACKUP_PART_NO=5 # in base_bullseye.yaml, instflux=0
else
BACKUP_PART_NO=6 # in base_bullseye.yaml, instflux=1 or not set
if [ "\${controller}" = true ] ; then
if [ "\${INSTFLUX}" = 0 ] ; then
BACKUP_PART_NO=5 # in base_bullseye.yaml, instflux=0
else
BACKUP_PART_NO=6 # in base_bullseye.yaml, instflux=1 or not set
fi
case \${INSTDEV} in
*"nvme"*)
BACKUP_PART=\${INSTDEV}p\${BACKUP_PART_NO}
;;
*)
BACKUP_PART=\${INSTDEV}\${BACKUP_PART_NO}
;;
esac
BACKUP_PART_LABEL="platform_backup"
DEFAULT_PERSISTENT_SIZE=30000
# Note that the BA5EBA11-0000-1111-2222- is the prefix used by STX and it's
# defined in sysinv constants.py. Since the 000000000001 suffix is used by
# custom stx LVM partitions, the next suffix is used for the persistent backup
# partition (000000000002)
BACKUP_PART_GUID="BA5EBA11-0000-1111-2222-000000000002"
fi
case \${INSTDEV} in
*"nvme"*)
BACKUP_PART=\${INSTDEV}p\${BACKUP_PART_NO}
;;
*)
BACKUP_PART=\${INSTDEV}\${BACKUP_PART_NO}
;;
esac
BACKUP_PART_LABEL="platform_backup"
DEFAULT_PERSISTENT_SIZE=30000
# Note that the BA5EBA11-0000-1111-2222- is the prefix used by STX and it's
# defined in sysinv constants.py. Since the 000000000001 suffix is used by
# custom stx LVM partitions, the next suffix is used for the persistent backup
# partition (000000000002)
BACKUP_PART_GUID="BA5EBA11-0000-1111-2222-000000000002"
# Node Personality Trait
TRAIT__CONTROLLER="controller"
@ -582,7 +633,6 @@ SUBFUNCTION__CONTROLLER="\${TRAIT__CONTROLLER}"
SUBFUNCTION__STORAGE="\${TRAIT__STORAGE}"
SUBFUNCTION__WORKER="\${TRAIT__WORKER}"
SUBFUNCTION__AIO="\${TRAIT__CONTROLLER},\${TRAIT__WORKER}"
SUBFUNCTION__AIO_LOWLATENCY="\${TRAIT__CONTROLLER},\${TRAIT__WORKER},\${TRAIT__LOWLATENCY}"
VOLUME_GROUPS="cgts-vg"
ROOTFS_SIZE=\${RSZ} # 20000
@ -591,6 +641,9 @@ EFI_SIZE=300
LOG_VOL_SIZE=8000
SCRATCH_VOL_SIZE=16000
EOF
###############################
# END of Common Functions Block
###############################
chmod 755 /tmp/lat/ks_functions.sh
HOOK_LABEL="ks-early"
@ -601,6 +654,7 @@ HOOK_LABEL="ks-early"
##########################################################################
export debug=0
export prefix=""
export PLATFORM_BACKUP_SIZE=${DEFAULT_PERSISTENT_SIZE}
@ -623,22 +677,20 @@ ilog "***********************************************"
ilog "** Pre Partition - Volume Remove & Wipe Disk **"
ilog "***********************************************"
if [ -z "${traits}" ] ; then
report_failure_with_msg "No install traits specified ; need boot arg traits=<controller|worker|storage|lowlatency>"
fi
ilog "Waiting for disks to be available"
udevadm settle --timeout=300 || report_failure_with_msg "udevadm settle failed"
ilog "Disks became available"
# Default traits to AIO if missing
if [ -z "${traits}" ] ; then
traits="controller,worker"
wlog "traits option missing ; defaulting to ${traits}"
fi
export controller=false
export storage=false
export worker=false
export lowlatency=false
export miniboot=false
export aio=false
for trait in ${traits//,/ }; do
# dlog "trait:${trait}"
case ${trait} in
@ -646,13 +698,10 @@ for trait in ${traits//,/ }; do
"storage") storage=true ;;
"worker") worker=true ;;
"lowlatency") lowlatency=true ;;
"miniboot") miniboot=true ;;
*) wlog "unknown trait '${trait}'" ;;
esac
done
ilog "Install traits: $traits"
# Force enable 'lowlatency' trait if the realtime kernel is detected ; PREEMPT_RT
uname=$(uname -a)
if [[ "${uname}" == *"PREEMPT_RT"* ]] ; then
@ -662,7 +711,38 @@ if [[ "${uname}" == *"PREEMPT_RT"* ]] ; then
fi
fi
[ "${controller}" = true -a "${worker}" = true ] && aio=true
# Handle trait conbination errors
#
# Handle no valid nodetypes specified
if [ "${controller}" = false -a "${storage}" = false -a "${worker}" = false ] ; then
report_failure_with_msg "No valid install traits specified ; ${traits} ; need boot arg traits=<controller|worker|storage>"
# Storage and controller don't go together
elif [ "${storage}" = true -a "${controller}" = true ] ; then
report_failure_with_msg "Cannot combine storage and controller traits"
# Storage and lowlatency don't go together
elif [ "${storage}" = true -a "${lowlatency}" = true ] ; then
report_failure_with_msg "Cannot combine storage and lowlatency traits"
# Storage and worker don't go together
elif [ "${storage}" = true -a "${worker}" = true ] ; then
report_failure_with_msg "Cannot combine storage and worker traits"
else
ilog "Install traits: $traits"
fi
# setup the install type for the log prefix
if [ "${controller}" = true -a "${worker}" = true ] ; then
aio=true
prefix="${SYSTEM_TYPE__AIO}"
elif [ "${controller}" = true ] ; then
prefix="${TRAIT__CONTROLLER}"
elif [ "${worker}" = true ] ; then
prefix="${TRAIT__WORKER}"
elif [ "${storage}" = true ] ; then
prefix="${TRAIT__STORAGE}"
fi
if [ "${lowlatency}" = true ] ; then
prefix="${prefix}-${TRAIT__LOWLATENCY}"
fi
grep -q 'debug_kickstart' /proc/cmdline
if [ $? -eq 0 ]; then
@ -710,11 +790,10 @@ function check_execs()
[ "${storage}" = true ] && ilog "Storage Function"
[ "${worker}" = true ] && ilog "Worker Function"
[ "${lowlatency}" = true ] && ilog "Low Latency Function"
[ "${miniboot}" = true ] && ilog "Zero Touch Install Function"
# Log System Type
if [ ${aio} = true ] ; then
ilog "${SYSTEM_TYPE__AIO} system"
ilog "system"
else
ilog "${SYSTEM_TYPE__STANDARD} system"
fi
@ -737,6 +816,39 @@ true
HOOK_LABEL="pre-part"
. /tmp/lat/ks_functions.sh
#####################################################
# From pre_net_common.cfg for system node installs
#####################################################
if is_system_node_install -eq 0 ; then
# Setup ntp.conf and sync time
cat <<EOF >/etc/ntp_kickstart.conf
server pxecontroller
EOF
if [ -e /usr/sbin/ntpd ] ; then
/usr/sbin/ntpd -g -q -n -c /etc/ntp_kickstart.conf
rc=$?
if [ ${rc} -eq 0 ]; then
if [ -e /sbin/hwclock ] ; then
/sbin/hwclock --systohc --utc
if [ $? -ne 0 ]; then
wlog "failed hwclock command ; /sbin/hwclock --systohc --utc"
else
ilog "ntpd and hwclock set ok"
fi
else
wlog "lat initrd is missing /sbin/hwclock"
fi
else
wlog "ntpd setup failed rc:$rc ; /usr/sbin/ntpd -g -q -n -c /etc/ntp_kickstart.conf"
fi
else
wlog "lat initrd is missing /usr/sbin/ntpd"
fi
fi
#####################################################
# From pre_disk_setup_common.cfg
#####################################################
@ -823,7 +935,7 @@ if [ -z ${vgs} ] ; then
else
ilog "Found '${vgs}' volume groups"
for vg in $vgs; do
wlog "... disabling $vg"
ilog "... disabling $vg"
exec_no_fds "$STOR_DEV_FDS" "vgchange -an $vg 2>/dev/null" 5 0.5
[ $? -ne 0 ] && report_failure_with_msg "Failed to disable $vg."
done
@ -835,12 +947,12 @@ else
wlog "Found $vg with no PV on rootfs, ignoring."
continue
fi
wlog "Removing LVs on $vg."
ilog "Removing LVs on $vg."
exec_no_fds "$STOR_DEV_FDS" "lvremove --force $vg" 5 0.5 || wlog "WARNING: Failed to remove lvs on $vg."
pvs=$(exec_no_fds "$STOR_DEV_FDS" "pvs --select \"vg_name=$vg\" --noheadings -o pv_name")
wlog "VG $vg has PVs: $(echo $pvs), removing them."
for pv in $pvs; do
wlog "Removing PV $pv."
ilog "Removing PV $pv."
exec_no_fds "$STOR_DEV_FDS" "pvremove --force --force --yes $pv" 5 0.5
[ $? -ne 0 ] && report_pre_failure_with_msg "ERROR: Failed to remove PV."
done
@ -885,10 +997,10 @@ else
# Check if we wipe OSDs
if [ "$(curl -sf http://pxecontroller:6385/v1/ihosts/wipe_osds 2>/dev/null)" = "true" ]; then
wlog "Wipe OSD data"
ilog "Wipe OSD data"
WIPE_CEPH_OSDS="true"
else
wlog "Skip Ceph OSD data wipe."
ilog "Skip Ceph OSD data wipe."
WIPE_CEPH_OSDS="false"
fi
@ -1042,34 +1154,34 @@ do
;;
esac
# ISSUE: ERIK: These are never the same when the by-path INSTDEV is used.
# dev = /dev/sda INSTDEV = by path string
ilog "Checking platform-backup partition ... $dev : ${INSTDEV} ${by_dev}"
sgdisk_part_info=$(sgdisk -i $part_number $dev)
part_type_guid=$(echo "$sgdisk_part_info" | grep "$part_type_guid_str" | awk '{print $4;}')
if [ "$dev" == "${by_dev}" -a "$part_type_guid" == $BACKUP_PART_GUID ] ; then
part_type_name=$(echo "$sgdisk_part_info" | grep "$part_type_name_str" | awk '{print $3,$4;}')
BACKUP_PART_NAME=${part_type_name:1:-1}
if [ "${controller}" = true ] ; then
ilog "Checking platform-backup partition ... $dev : ${INSTDEV} ${by_dev}"
sgdisk_part_info=$(sgdisk -i $part_number $dev)
part_type_guid=$(echo "$sgdisk_part_info" | grep "$part_type_guid_str" | awk '{print $4;}')
if [ "$dev" == "${by_dev}" -a "$part_type_guid" == $BACKUP_PART_GUID ] ; then
part_type_name=$(echo "$sgdisk_part_info" | grep "$part_type_name_str" | awk '{print $3,$4;}')
BACKUP_PART_NAME=${part_type_name:1:-1}
part_type_first=$(echo "$sgdisk_part_info" | grep "$part_type_first_str" | awk '{print $3;}')
BACKUP_PART_FIRST=${part_type_first}
part_type_first=$(echo "$sgdisk_part_info" | grep "$part_type_first_str" | awk '{print $3;}')
BACKUP_PART_FIRST=${part_type_first}
part_type_end=$(echo "$sgdisk_part_info" | grep "$part_type_end_str" | awk '{print $3;}')
BACKUP_PART_END=${part_type_end}
part_type_end=$(echo "$sgdisk_part_info" | grep "$part_type_end_str" | awk '{print $3;}')
BACKUP_PART_END=${part_type_end}
part_type_size=$(echo "$sgdisk_part_info" | grep "$part_type_size_str" | awk '{print $3;}')
BACKUP_PART_SIZE=${part_type_size}
part_type_size=$(echo "$sgdisk_part_info" | grep "$part_type_size_str" | awk '{print $3;}')
BACKUP_PART_SIZE=${part_type_size}
part_type_flags=$(echo "$sgdisk_part_info" | grep "$part_type_flags_str" | awk '{print $3;}')
BACKUP_PART_FLAGS=${part_type_flags}
part_type_flags=$(echo "$sgdisk_part_info" | grep "$part_type_flags_str" | awk '{print $3;}')
BACKUP_PART_FLAGS=${part_type_flags}
ilog "Platform Backup: ${BACKUP_PART_NAME}:${BACKUP_PART_FIRST}:${BACKUP_PART_END}:${BACKUP_PART_SIZE}:${part_type_guid}"
ilog "Platform Backup: ${BACKUP_PART_NAME}:${BACKUP_PART_FIRST}:${BACKUP_PART_END}:${BACKUP_PART_SIZE}:${part_type_guid}"
part_fstype=$(exec_retry 5 0.5 "blkid -s TYPE -o value $part")
if [ "${part_fstype}" == "ext4" ]; then
ilog "Skipping wipe of persistent partition $part"
BACKUP_CREATED=1
continue
part_fstype=$(exec_retry 5 0.5 "blkid -s TYPE -o value $part")
if [ "${part_fstype}" == "ext4" ]; then
ilog "Skipping wipe of persistent partition $part"
BACKUP_CREATED=1
continue
fi
fi
fi
@ -1103,7 +1215,7 @@ for dev in $STOR_DEVS; do
fi
done
wlog "Destroy otaroot partitions on root disk"
ilog "Destroy otaroot partitions on root disk"
for oldrootlabel in otaroot otaroot_1 otaroot_b otaroot_b_1
do
oldrootpart=$(blkid --label $oldrootlabel)
@ -1160,6 +1272,13 @@ esac
# From pre_disk_controller.cfg
###########################################################
# Log the server firmware type
if [ -d /sys/firmware/efi ] ; then
ilog "UEFI Firmware"
else
ilog "BIOS Firmware"
fi
# Only controllers have a persistent partition
# Allocate platform backup partition
if [ "${controller}" = true ] ; then
@ -1179,12 +1298,6 @@ if [ "${controller}" = true ] ; then
fi
ilog "Platform Backup partition size: ${PLATFORM_BACKUP_SIZE}"
if [ -d /sys/firmware/efi ] ; then
ilog "UEFI Firmware"
else
ilog "BIOS Firmware"
fi
if [ ${BACKUP_CREATED} -eq 0 ] ; then
ilog "Platform Backup partition not detected: CREATING"
else
@ -1200,6 +1313,10 @@ if [ "${controller}" = true ] ; then
grub_pt_update
end=$(($first+($PLATFORM_BACKUP_SIZE*1024*1024/$lsz)-1))
a="$a -n $p:$first:$end -c $p:${BACKUP_PART_LABEL} -t $p:${BACKUP_PART_GUID}"
elif [ ${worker} = true ] ; then
ilog "Setting up platform_pv partition sizes for worker node"
get_worker_provisioning_sizes
fi
# Allocate platform pv
@ -1207,7 +1324,7 @@ grub_pt_update
if [ ${aio} = true ] ; then
end=$(($first+(${MINIMUM_PLATFORM_PV_SIZE}*1024*1024/$lsz)-1))
else
ilog "Standard Controller: platform_pv partition - using all remaining disk space"
ilog "platform_pv partition - using all remaining disk space"
end=$last
fi
a="$a -n $p:$first:$end -c $p:platform_pv -t $p:8E00"
@ -1240,6 +1357,10 @@ ilog "****************************************************"
ilog "LAT size parameters: FSZ=${FSZ} BSZ=${BSZ} RSZ=${RSZ} VSZ=${VSZ} gb:${gb}"
vg="volume group"
lv="logical volume"
dev=$(get_disk "${INSTDEV}")
if [ "${controller}" = true ] ; then
LOG_VOL_SIZE=8000
SCRATCH_VOL_SIZE=16000
@ -1256,29 +1377,29 @@ if [ "${aio}" = true ] ; then
get_aio_controller_provisioning_sizes
elif [ "${controller}" = true ] ; then
get_std_controller_provisioning_sizes
elif [ "${worker}" = true ] ; then
get_worker_provisioning_sizes
fi
ilog "MINIMUM_PLATFORM_PV_SIZE : ${MINIMUM_PLATFORM_PV_SIZE}"
ilog "SCRATCH_VOL_SIZE : ${SCRATCH_VOL_SIZE}"
ilog "ROOTFS_SIZE : ${ROOTFS_SIZE}"
ilog "LOG_VOL_SIZE : ${LOG_VOL_SIZE}"
ilog "PLATFORM_BACKUP_SIZE : ${PLATFORM_BACKUP_SIZE}"
ilog "BACKUP_PART : ${BACKUP_PART}"
ilog "BACKUP_PART_NO : ${BACKUP_PART_NO}"
ilog "BACKUP_PART_LABEL : ${BACKUP_PART_LABEL}"
ilog "BACKUP_PART_GUID : ${BACKUP_PART_GUID}"
if [ "${controller}" = true ] ; then
ilog "PLATFORM_BACKUP_SIZE : ${PLATFORM_BACKUP_SIZE}"
ilog "BACKUP_PART : ${BACKUP_PART}"
ilog "BACKUP_PART_NO : ${BACKUP_PART_NO}"
ilog "BACKUP_PART_LABEL : ${BACKUP_PART_LABEL}"
ilog "BACKUP_PART_GUID : ${BACKUP_PART_GUID}"
vg="volume group"
lv="logical volume"
dev=$(get_disk "${INSTDEV}")
# Only init Platform Backup partition filesystem if the partition was just created
pi=$((pi+1))
if [ ${BACKUP_PERSISTED} -eq 0 ] ; then
ilog "Platform Backup Partition was CREATED. Initialize filesystem on ${BACKUP_PART}"
mkfs.ext4 -F -L ${BACKUP_PART_LABEL} ${BACKUP_PART}
[ ${?} -ne 0 ] && report_failure_with_msg "Failed Platform Backup partition filesystem init ${BACKUP_PART}"
# Only init Platform Backup partition filesystem if the partition was just created
pi=$((pi+1))
if [ ${BACKUP_PERSISTED} -eq 0 ] ; then
ilog "Platform Backup Partition was CREATED. Initialize filesystem on ${BACKUP_PART}"
mkfs.ext4 -F -L ${BACKUP_PART_LABEL} ${BACKUP_PART}
[ ${?} -ne 0 ] && report_failure_with_msg "Failed Platform Backup partition filesystem init ${BACKUP_PART}"
fi
fi
# Log important LVM config settings
@ -1348,7 +1469,7 @@ log_lvm_conf
#####################################################################
# Close all FDs and wait for udev to reshuffle all partitions.
wlog "Releasing storage device locks and FDs."
ilog "Releasing storage device locks and FDs."
for fd in $STOR_DEV_FDS
do
flock -u "$fd"
@ -1374,21 +1495,6 @@ ilog "****************************************************"
ilog "**** Post - Traits & Platform Config File update **"
ilog "****************************************************"
set -- `env`
dlog "env: $*"
if false ; then
# TODO: Why does LAT break into prompt when I do this
for e in $*; do
case "$e" in
*=*)
eval $e 2>/dev/null
log "env : $e"
;;
esac
done
fi
# Set node traits' state
[ "${controller}" = true ] && ilog "Controller Function"
[ "${storage}" = true ] && ilog "Storage Function"
@ -1402,17 +1508,10 @@ fi
# Set nodetype from traits
if [ "${controller}" = true ] ; then
nodetype=${TRAIT__CONTROLLER}
ilog "Controller Function"
elif [ "${storage}" = true ] ; then
nodetype=${TRAIT__STORAGE}
ilog "Storage Function"
elif [ "${worker}" = true ] ; then
nodetype=${TRAIT__WORKER}
else
elog "No personality traits specified ; defaulting to All-In-One"
controller=true
worker=true
nodetype=${TRAIT__CONTROLLER}
fi
ilog "nodetype=${nodetype}"
@ -1423,11 +1522,7 @@ system_type=""
if [ "${aio}" = true ] ; then
ilog "${SYSTEM_TYPE__AIO} System"
system_type="${SYSTEM_TYPE__AIO}"
if [ "${lowlatency}" = true ] ; then
subfunction="${SUBFUNCTION__AIO_LOWLATENCY}"
else
subfunction="${SUBFUNCTION__AIO}"
fi
subfunction="${SUBFUNCTION__AIO}"
else
ilog "${SYSTEM_TYPE__STANDARD} System"
system_type="${SYSTEM_TYPE__STANDARD}"
@ -1439,16 +1534,14 @@ else
subfunction=${SUBFUNCTION__CONTROLLER}
fi
fi
if [ "${lowlatency}" = true ] ; then
subfunction="${subfunction},${TRAIT__LOWLATENCY}"
fi
ilog "subfunction=${subfunction}"
ilog "system_type=${system_type}"
# Create and Update /etc/platform/platform.conf
# ---------------------------------------------
if [ -e "${PLATFORM_CONF}" ] ; then
wlog "removing temp platform.conf"
fi
#################################################################
# From post_platform_conf_controller.cfg
# From post_platform_conf_aio.cfg
@ -1463,6 +1556,10 @@ subfunction=${subfunction}
system_type=${system_type}
EOF
##############################################################
# From post_pxeboot_controller
# From post_net_common.cfg
# ############################################################
# Set http port
# Persist the default http port number to platform configuration.
# This will get overwritten when by manifest during sw configuration phase.
@ -1477,7 +1574,6 @@ update_platform_conf "http_port=${http_port}"
# From post_common.cfg (all hosts)
#############################################################
# Ensure quick command response (i.e. sudo)
rm /etc/hostname
# TODO: is this needed with LAT ??
@ -1548,18 +1644,11 @@ do
fi
done
# TODO: Remove as it turns out that this is not added by kickstarts.
# Remove all reference to 'security_feature' above as well.
#
# add security_feature to platform.conf ; be sure not to add as empty.
# [ -z "${security_feature}" ] && security_feature=none
# update_platform_conf "security_feature=${security_feature}"
[ "${KERN_OPTS}" != "" ] && ilog "Learned kernel params: ${KERN_OPTS}"
if [ "${aio}" = true ] ; then
ilog "Adding kernel options"
if [ "${aio}" = true -o "${worker}" = true ] ; then
ilog "Adding ${SYSTEM_TYPE__AIO} kernel options"
if [ -e /etc/init.d/cpumap_functions.sh ] ; then
# Update grub with custom kernel bootargs
@ -1629,13 +1718,6 @@ if [ "${aio}" = true ] ; then
add_kernel_option "intel_iommu=on"
fi
# Add kernel option to disable biosdevname if enabled
# As this may already be in GRUB_CMDLINE_LINUX, only add if it is not already present
grep -q '^GRUB_CMDLINE_LINUX=.*biosdevname=0' /etc/default/grub
if [ $? -ne 0 ]; then
add_kernel_option "biosdevname=0"
fi
# Add kernel options to disable kvm-intel.eptad on Broadwell
# Broadwell: Model: 79, Model name: Intel(R) Xeon(R) CPU E5-2699 v4 @ 2.20GHz
if grep -q -E "^model\s+:\s+79$" /proc/cpuinfo
@ -1643,10 +1725,6 @@ if [ "${aio}" = true ] ; then
add_kernel_option "kvm-intel.eptad=0"
fi
# k8s updates:
#KERN_OPTS="${KERN_OPTS} cgroup_disable=memory"
add_kernel_option "user_namespace.enable=1"
# Add kernel option to avoid jiffies_lock contention on real-time kernel
if [ "${lowlatency}" = true ] ; then
add_kernel_option "skew_tick=1"
@ -1664,29 +1742,48 @@ elif [ "${controller}" = true ] ; then
##
add_kernel_option "loop.max_part=15"
## Add kernel options to ensure an selinux is disabled
add_kernel_option "selinux=0 enforcing=0"
# Add kernel options to ensure NMI watchdog is enabled, if supported
add_kernel_option "nmi_watchdog=panic,1 softlockup_panic=1"
# Add kernel option to panic on a softdog timeout
add_kernel_option "softdog.soft_panic=1"
elif [ "${storage}" = true ] ; then
# Add kernel option to disable biosdevname if enabled
# As this may already be in GRUB_CMDLINE_LINUX, only add if it is not already present
grep -q '^GRUB_CMDLINE_LINUX=.*biosdevname=0' /etc/default/grub
[ $? -ne 0 ] && add_kernel_option "biosdevname=0"
ilog "Adding ${TRAIT__STORAGE} function kernel options"
# k8s updates
add_kernel_option "user_namespace.enable=1"
else
wlog "TODO: system node install type kernel parameters, worker and storage"
## Custom kernel options
add_kernel_option "intel_iommu=off usbcore.autosuspend=-1"
# Add kernel options to ensure NMI watchdog is enabled, if supported
add_kernel_option "nmi_watchdog=panic,1 softlockup_panic=1"
fi
#########################################################################################
# From post_kernel_aio_and_worker.cfg
# From post_kernel_controller.cfg
# From post_kernel_storage.cfg
#########################################################################################
# Add kernel options to ensure an selinux is disabled for all node types
add_kernel_option "selinux=0 enforcing=0"
# Add kernel option to panic on a softdog timeout
add_kernel_option "softdog.soft_panic=1"
# Make sure that we are using cgroups v1
add_kernel_option "systemd.unified_cgroup_hierarchy=0"
# k8s updates
add_kernel_option "user_namespace.enable=1"
# Add kernel option to disable biosdevname if enabled
# As this may already be in GRUB_CMDLINE_LINUX, only add if it is not already present
grep -q '^GRUB_CMDLINE_LINUX=.*biosdevname=0' /etc/default/grub
if [ $? -ne 0 ]; then
add_kernel_option "biosdevname=0"
fi
########################################################################################
ilog "Adding these kernel params to disk boot: ${KERN_OPTS}"
if [ -n "${KERN_OPTS}" ] ; then
grep -q "^kernel_params=" /boot/efi/EFI/BOOT/boot.env
@ -1792,12 +1889,10 @@ if [ -n "$BOOTIF" ] ; then
fi
else
elog "ERROR:POST: Unable to determine mgmt interface from BOOTIF=$BOOTIF."
# TODO: MAKE Default - Should not get here without BOOTIF set ; LAT guarantees that.
# report_failure_with_msg "Unable to determine mgmt interface from BOOTIF=$BOOTIF."
report_failure_with_msg "Unable to determine mgmt interface from BOOTIF=$BOOTIF."
fi
else
elog "ERROR:POST: BOOTIF is not set. Unable to determine mgmt interface."
# TODO: MAKE Default - Should not get here without BOOTIF set ; LAT guarantees that.
wlog "BOOTIF is not set. Unable to determine mgmt interface."
fi
if [ ! -e "/etc/network/interfaces" ] ; then
@ -1926,6 +2021,10 @@ else
else
# must be a worker or storage node install
pushd /tmp > /dev/null
# There is no feed function on non-controller nodes
# TODO: Prevent feed dir packaging on non-controller nodes
rm -rf ${IMAGE_ROOTFS}/var/www/pages/feed
fi
if [ ! -e ./install_uuid ] ; then
wget ${NOVERIFYSSL_WGET_OPT} ${pxeurl}/install_uuid -o /${LAT_DIR}/wget_install_uuid.log
@ -2020,9 +2119,7 @@ if [ "${controller}" = true ] ; then
fi
# handle USB install
pxeurl=$(echo $insturl | sed -e s/ostree_repo//)
if [ "${pxeurl}" = "file://NOT_SET" ] ; then
if is_usb_install -eq 0 ; then
pushd ${feed}/pxeboot > /dev/null
ilog "Copy bzImage, initrd and respective secure boot .sig files"
ilog "... from /instboot/pxeboot to ${feed}/pxeboot and ${pxeboot}/rel-${sw_release}"
@ -2045,6 +2142,7 @@ if [ "${controller}" = true ] ; then
# handle pxeboot install
else
pxeurl=$(echo $insturl | sed -e s/ostree_repo//)
ilog "Fetch bzImage and initrd files from ${pxeurl}/pxeboot to ${feed}/pxeboot and ${pxeboot}/rel-${sw_release}"
pushd ${feed}/pxeboot > /dev/null
for f in bzImage bzImage-rt bzImage-rt.sig bzImage.sig bzImage-std bzImage-std.sig initrd initrd.sig; do
@ -2162,11 +2260,6 @@ else
umount "${IMAGE_ROOTFS}/${LOG_DIR}"
fi
# uncomment the global_filter line in lvm.conf
#perl -0777 -i.bak -pe 's:(# This configuration option has an automatic default value\.\n)\t# global_filter:$1 global_filter:m' ${IMAGE_ROOTFS}/etc/lvm/lvm.conf
#perl -0777 -i -pe 's:(# Example\n)\t# preferred_names:$1 preferred_names:m' ${IMAGE_ROOTFS}/etc/lvm/lvm.conf
#sed -i "s#^\( *\)global_filter = \[.*#\1global_filter = [ \"a|${LV_ROOTDISK}|\", \"r|.*|\" ]#" ${IMAGE_ROOTFS}/etc/lvm/lvm.conf
true
%end
@ -2182,11 +2275,18 @@ ilog "****************************************************"
ilog "*** Post Nochroot - Set up /etc/fstab ***"
ilog "****************************************************"
mkdir -p ${IMAGE_ROOTFS}/var/rootdirs/opt/platform-backup
if [ ${controller} = true ] ; then
mkdir -p ${IMAGE_ROOTFS}/var/rootdirs/opt/platform-backup
echo -e "LABEL=platform_backup\t/var/rootdirs/opt/platform-backup\text4\tdefaults\t1 2" >> ${IMAGE_ROOTFS}/etc/fstab
elif [ ${worker} = true ] ; then
mkdir -p ${IMAGE_ROOTFS}/var/rootdir/opt/platform
echo -e "controller-platform-nfs:/opt/platform\t/opt/platform\tnfs\ttimeo=30,udp,rsize=1024,wsize=1024,_netdev 0 0" >> ${IMAGE_ROOTFS}/etc/fstab
fi
mkdir -p ${IMAGE_ROOTFS}/var/rootdirs/scratch
mkdir -p ${IMAGE_ROOTFS}/var/log
echo -e "LABEL=platform_backup\t/var/rootdirs/opt/platform-backup\text4\tdefaults\t1 2" >> ${IMAGE_ROOTFS}/etc/fstab
echo -e "/dev/mapper/cgts--vg-scratch--lv\t/var/rootdirs/scratch\text4\tdefaults\t1 2" >> ${IMAGE_ROOTFS}/etc/fstab
mkdir -p ${IMAGE_ROOTFS}/var/log
echo -e "/dev/mapper/cgts--vg-log--lv\t/var/log\text4\tdefaults\t1 2" >> ${IMAGE_ROOTFS}/etc/fstab
# Create first_boot flag