diff --git a/files/apts/baremetal b/files/apts/baremetal new file mode 100644 index 0000000000..54e76e0064 --- /dev/null +++ b/files/apts/baremetal @@ -0,0 +1,9 @@ +busybox +dnsmasq +gcc +ipmitool +make +open-iscsi +qemu-kvm +syslinux +tgt diff --git a/lib/baremetal b/lib/baremetal new file mode 100644 index 0000000000..f82633a409 --- /dev/null +++ b/lib/baremetal @@ -0,0 +1,403 @@ +# vim: tabstop=4 shiftwidth=4 softtabstop=4 + +# Copyright (c) 2012 Hewlett-Packard Development Company, L.P. +# All Rights Reserved. +# +# Licensed under the Apache License, Version 2.0 (the "License"); you may +# not use this file except in compliance with the License. You may obtain +# a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, WITHOUT +# WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the +# License for the specific language governing permissions and limitations +# under the License. + + +# This file provides devstack with the environment and utilities to +# control nova-compute's baremetal driver. +# It sets reasonable defaults to run within a single host, +# using virtual machines in place of physical hardware. +# However, by changing just a few options, devstack+baremetal can in fact +# control physical hardware resources on the same network, if you know +# the MAC address(es) and IPMI credentials. +# +# At a minimum, to enable the baremetal driver, you must set these in loclarc: +# VIRT_DRIVER=baremetal +# ENABLED_SERVICES="$ENABLED_SERVICES,baremetal" +# +# +# We utilize diskimage-builder to create a ramdisk, and then +# baremetal driver uses that to push a disk image onto the node(s). +# +# Below we define various defaults which control the behavior of the +# baremetal compute service, and inform it of the hardware it will contorl. +# +# Below that, various functions are defined, which are called by devstack +# in the following order: +# +# before nova-cpu starts: +# - prepare_baremetal_toolchain +# - configure_baremetal_nova_dirs +# +# after nova and glance have started: +# - build_and_upload_baremetal_deploy_k_and_r $token +# - create_baremetal_flavor $BM_DEPLOY_KERNEL_ID $BM_DEPLOY_RAMDISK_ID +# - upload_baremetal_image $url $token +# - add_baremetal_node + + +# Save trace setting +XTRACE=$(set +o | grep xtrace) +set +o xtrace + +# Sub-driver settings +# ------------------- + +# sub-driver to use for kernel deployment +# - nova.virt.baremetal.pxe.PXE +# - nova.virt.baremetal.tilera.TILERA +BM_DRIVER=${BM_DRIVER:-nova.virt.baremetal.pxe.PXE} + +# sub-driver to use for remote power management +# - nova.virt.baremetal.fake.FakePowerManager, for manual power control +# - nova.virt.baremetal.ipmi.Ipmi, for remote IPMI +# - nova.virt.baremetal.tilera_pdu.Pdu, for TilePro hardware +BM_POWER_MANAGER=${BM_POWER_MANAGER:-nova.virt.baremetal.fake.FakePowerManager} + + +# These should be customized to your environment and hardware +# ----------------------------------------------------------- + +# BM_DNSMASQ_* options must be changed to suit your network environment +BM_DNSMASQ_IFACE=${BM_DNSMASQ_IFACE:-$PUBLIC_INTERFACE} +BM_DNSMASQ_IFACE=${BM_DNSMASQ_IFACE:-eth0} +BM_DNSMASQ_RANGE=${BM_DNSMASQ_RANGE:-192.0.2.32,192.0.2.48} + +# BM_FIRST_MAC *must* be set to the MAC address of the node you will boot. +# This is passed to dnsmasq along with the kernel/ramdisk to +# deploy via PXE. +BM_FIRST_MAC=${BM_FIRST_MAC:-} + +# BM_SECOND_MAC is only important if the host has >1 NIC. +BM_SECOND_MAC=${BM_SECOND_MAC:-} + +# Hostname for the baremetal nova-compute node, if not run on this host +BM_HOSTNAME=${BM_HOSTNAME:-$(hostname -f)} + +# BM_PM_* options are only necessary if BM_POWER_MANAGER=...IPMI +BM_PM_ADDR=${BM_PM_ADDR:-0.0.0.0} +BM_PM_USER=${BM_PM_USER:-user} +BM_PM_PASS=${BM_PM_PASS:-pass} + +# BM_FLAVOR_* options are arbitrary and not necessarily related to physical +# hardware capacity. These can be changed if you are testing +# BaremetalHostManager with multiple nodes and different flavors. +BM_CPU_ARCH=${BM_CPU_ARCH:-x86_64} +BM_FLAVOR_CPU=${BM_FLAVOR_CPU:-1} +BM_FLAVOR_RAM=${BM_FLAVOR_RAM:-1024} +BM_FLAVOR_ROOT_DISK=${BM_FLAVOR_ROOT_DISK:-10} +BM_FLAVOR_EPHEMERAL_DISK=${BM_FLAVOR_EPHEMERAL_DISK:-0} +BM_FLAVOR_SWAP=${BM_FLAVOR_SWAP:-1} +BM_FLAVOR_NAME=${BM_FLAVOR_NAME:-bm.small} +BM_FLAVOR_ID=${BM_FLAVOR_ID:-11} +BM_FLAVOR_ARCH=${BM_FLAVOR_ARCH:-$BM_CPU_ARCH} + + +# Below this, we set some path and filenames. +# Defaults are probably sufficient. + +BM_IMAGE_BUILD_DIR=${BM_IMAGE_BUILD_DIR:-$DEST/diskimage-builder} + +BM_HOST_CURRENT_KERNEL=$(uname -r) +BM_DEPLOY_RAMDISK=${BM_DEPLOY_RAMDISK:-bm-deploy-$BM_HOST_CURRENT_KERNEL-initrd} +BM_DEPLOY_KERNEL=${BM_DEPLOY_KERNEL:-bm-deploy-$BM_HOST_CURRENT_KERNEL-vmlinuz} + +# If you need to add any extra flavors to the deploy ramdisk image +# eg, specific network drivers, specify them here +BM_DEPLOY_FLAVOR=${BM_DEPLOY_FLAVOR:-} + +# set URL and version for google shell-in-a-box +BM_SHELL_IN_A_BOX=${BM_SHELL_IN_A_BOX:-http://shellinabox.googlecode.com/files/shellinabox-2.14.tar.gz} + + +# Functions +# --------- + +# Check if baremetal is properly enabled +# Returns false if VIRT_DRIVER is not baremetal, or if ENABLED_SERVICES +# does not contain "baremetal" +function is_baremetal() { + if [[ "$ENABLED_SERVICES" =~ 'baremetal' && "$VIRT_DRIVER" = 'baremetal' ]]; then + return 0 + fi + return 1 +} + +# Install diskimage-builder and shell-in-a-box +# so that we can build the deployment kernel & ramdisk +function prepare_baremetal_toolchain() { + git_clone $BM_IMAGE_BUILD_REPO $BM_IMAGE_BUILD_DIR $BM_IMAGE_BUILD_BRANCH + + local shellinabox_basename=$(basename $BM_SHELL_IN_A_BOX) + if [[ ! -e $DEST/$shellinabox_basename ]]; then + cd $DEST + wget $BM_SHELL_IN_A_BOX + fi + if [[ ! -d $DEST/${shellinabox_basename%%.tar.gz} ]]; then + cd $DEST + tar xzf $shellinabox_basename + fi + if [[ ! $(which shellinaboxd) ]]; then + cd $DEST/${shellinabox_basename%%.tar.gz} + ./configure + make + sudo make install + fi +} + +# prepare various directories needed by baremetal hypervisor +function configure_baremetal_nova_dirs() { + # ensure /tftpboot is prepared + sudo mkdir -p /tftpboot + sudo mkdir -p /tftpboot/pxelinux.cfg + sudo cp /usr/lib/syslinux/pxelinux.0 /tftpboot/ + sudo chown -R `whoami`:libvirtd /tftpboot + + # ensure $NOVA_STATE_PATH/baremetal is prepared + sudo mkdir -p $NOVA_STATE_PATH/baremetal + sudo mkdir -p $NOVA_STATE_PATH/baremetal/console + sudo mkdir -p $NOVA_STATE_PATH/baremetal/dnsmasq + sudo touch $NOVA_STATE_PATH/baremetal/dnsmasq/dnsmasq-dhcp.host + sudo chown -R `whoami` $NOVA_STATE_PATH/baremetal + + # ensure dnsmasq is installed but not running + # because baremetal driver will reconfigure and restart this as needed + if [ ! is_package_installed dnsmasq ]; then + install_package dnsmasq + fi + stop_service dnsmasq +} + +# build deploy kernel+ramdisk, then upload them to glance +# this function sets BM_DEPLOY_KERNEL_ID and BM_DEPLOY_RAMDISK_ID +function upload_baremetal_deploy() { + token=$1 + + if [ ! -e $TOP_DIR/files/$BM_DEPLOY_KERNEL -a -e /boot/vmlinuz-$BM_HOST_CURRENT_KERNEL ]; then + sudo cp /boot/vmlinuz-$BM_HOST_CURRENT_KERNEL $TOP_DIR/files/$BM_DEPLOY_KERNEL + sudo chmod a+r $TOP_DIR/files/$BM_DEPLOY_KERNEL + fi + if [ ! -e $TOP_DIR/files/$BM_DEPLOY_RAMDISK ]; then + $BM_IMAGE_BUILD_DIR/bin/ramdisk-image-create $BM_DEPLOY_FLAVOR deploy \ + -o $TOP_DIR/files/$BM_DEPLOY_RAMDISK -k $BM_HOST_CURRENT_KERNEL + fi + + # load them into glance + BM_DEPLOY_KERNEL_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name $BM_DEPLOY_KERNEL \ + --public --disk-format=aki \ + < $TOP_DIR/files/$BM_DEPLOY_KERNEL | grep ' id ' | get_field 2) + BM_DEPLOY_RAMDISK_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name $BM_DEPLOY_RAMDISK \ + --public --disk-format=ari \ + < $TOP_DIR/files/$BM_DEPLOY_RAMDISK | grep ' id ' | get_field 2) +} + +# create a basic baremetal flavor, associated with deploy kernel & ramdisk +# +# Usage: create_baremetal_flavor +function create_baremetal_flavor() { + aki=$1 + ari=$2 + nova flavor-create $BM_FLAVOR_NAME $BM_FLAVOR_ID \ + $BM_FLAVOR_RAM $BM_FLAVOR_ROOT_DISK $BM_FLAVOR_CPU + nova-manage instance_type set_key \ + --name=$BM_FLAVOR_NAME --key cpu_arch --value $BM_FLAVOR_ARCH + nova-manage instance_type set_key \ + --name=$BM_FLAVOR_NAME --key deploy_kernel_id --value $aki + nova-manage instance_type set_key \ + --name=$BM_FLAVOR_NAME --key deploy_ramdisk_id --value $ari +} + +# pull run-time kernel/ramdisk out of disk image and load into glance +# note that $file is currently expected to be in qcow2 format +# Sets KERNEL_ID and RAMDISK_ID +# +# Usage: extract_and_upload_k_and_r_from_image $token $file +function extract_and_upload_k_and_r_from_image() { + token=$1 + file=$2 + image_name=$(basename "$file" ".qcow2") + + # this call returns the file names as "$kernel,$ramdisk" + out=$($BM_IMAGE_BUILD_DIR/bin/disk-image-get-kernel \ + -x -d $TOP_DIR/files -o bm-deploy -i $file) + if [ $? -ne 0 ]; then + die "Failed to get kernel and ramdisk from $file" + fi + XTRACE=$(set +o | grep xtrace) + set +o xtrace + out=$(echo "$out" | tail -1) + $XTRACE + OUT_KERNEL=${out%%,*} + OUT_RAMDISK=${out##*,} + + # load them into glance + KERNEL_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name $image_name-kernel \ + --public --disk-format=aki \ + < $TOP_DIR/files/$OUT_KERNEL | grep ' id ' | get_field 2) + RAMDISK_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name $image_name-initrd \ + --public --disk-format=ari \ + < $TOP_DIR/files/$OUT_RAMDISK | grep ' id ' | get_field 2) +} + + +# Re-implementation of devstack's "upload_image" function +# +# Takes the same parameters, but has some peculiarities which made it +# easier to create a separate method, rather than complicate the logic +# of the existing function. +function upload_baremetal_image() { + local image_url=$1 + local token=$2 + + # Create a directory for the downloaded image tarballs. + mkdir -p $FILES/images + + # Downloads the image (uec ami+aki style), then extracts it. + IMAGE_FNAME=`basename "$image_url"` + if [[ ! -f $FILES/$IMAGE_FNAME || \ + "$(stat -c "%s" $FILES/$IMAGE_FNAME)" = "0" ]]; then + wget -c $image_url -O $FILES/$IMAGE_FNAME + if [[ $? -ne 0 ]]; then + echo "Not found: $image_url" + return + fi + fi + + local KERNEL="" + local RAMDISK="" + local DISK_FORMAT="" + local CONTAINER_FORMAT="" + case "$IMAGE_FNAME" in + *.tar.gz|*.tgz) + # Extract ami and aki files + [ "${IMAGE_FNAME%.tar.gz}" != "$IMAGE_FNAME" ] && + IMAGE_NAME="${IMAGE_FNAME%.tar.gz}" || + IMAGE_NAME="${IMAGE_FNAME%.tgz}" + xdir="$FILES/images/$IMAGE_NAME" + rm -Rf "$xdir"; + mkdir "$xdir" + tar -zxf $FILES/$IMAGE_FNAME -C "$xdir" + KERNEL=$(for f in "$xdir/"*-vmlinuz* "$xdir/"aki-*/image; do + [ -f "$f" ] && echo "$f" && break; done; true) + RAMDISK=$(for f in "$xdir/"*-initrd* "$xdir/"ari-*/image; do + [ -f "$f" ] && echo "$f" && break; done; true) + IMAGE=$(for f in "$xdir/"*.img "$xdir/"ami-*/image; do + [ -f "$f" ] && echo "$f" && break; done; true) + if [[ -z "$IMAGE_NAME" ]]; then + IMAGE_NAME=$(basename "$IMAGE" ".img") + fi + DISK_FORMAT=ami + CONTAINER_FORMAT=ami + ;; + *.qcow2) + IMAGE="$FILES/${IMAGE_FNAME}" + IMAGE_NAME=$(basename "$IMAGE" ".qcow2") + DISK_FORMAT=qcow2 + CONTAINER_FORMAT=bare + ;; + *) echo "Do not know what to do with $IMAGE_FNAME"; false;; + esac + + if [ "$CONTAINER_FORMAT" = "bare" ]; then + extract_and_upload_k_and_r_from_image $token $IMAGE + elif [ "$CONTAINER_FORMAT" = "ami" ]; then + KERNEL_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name "$IMAGE_NAME-kernel" --public \ + --container-format aki \ + --disk-format aki < "$KERNEL" | grep ' id ' | get_field 2) + RAMDISK_ID=$(glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name "$IMAGE_NAME-ramdisk" --public \ + --container-format ari \ + --disk-format ari < "$RAMDISK" | grep ' id ' | get_field 2) + else + # TODO(deva): add support for other image types + return + fi + + glance \ + --os-auth-token $token \ + --os-image-url http://$GLANCE_HOSTPORT \ + image-create \ + --name "${IMAGE_NAME%.img}" --public \ + --container-format $CONTAINER_FORMAT \ + --disk-format $DISK_FORMAT \ + ${KERNEL_ID:+--property kernel_id=$KERNEL_ID} \ + ${RAMDISK_ID:+--property ramdisk_id=$RAMDISK_ID} < "${IMAGE}" + + # override DEFAULT_IMAGE_NAME so that tempest can find the image + # that we just uploaded in glance + DEFAULT_IMAGE_NAME="${IMAGE_NAME%.img}" +} + +function clear_baremetal_of_all_nodes() { + list=$(nova-baremetal-manage node list | tail -n +2 | awk '{print $1}' ) + for node in $list + do + nova-baremetal-manage node delete $node + done + list=$(nova-baremetal-manage interface list | tail -n +2 | awk '{print $1}' ) + for iface in $list + do + nova-baremetal-manage interface delete $iface + done +} + +# inform nova-baremetal about nodes, MACs, etc +# Defaults to using BM_FIRST_MAC and BM_SECOND_MAC if parameters not specified +# +# Usage: add_baremetal_node +function add_baremetal_node() { + mac_1=${1:-$BM_FIRST_MAC} + mac_2=${2:-$BM_SECOND_MAC} + + id=$(nova-baremetal-manage node create \ + --host=$BM_HOSTNAME --prov_mac=$mac_1 \ + --cpus=$BM_FLAVOR_CPU --memory_mb=$BM_FLAVOR_RAM \ + --local_gb=$BM_FLAVOR_ROOT_DISK --terminal_port=0 \ + --pm_address=$BM_PM_ADDR --pm_user=$BM_PM_USER --pm_password=$BM_PM_PASS \ + ) + [ $? -eq 0 ] || [ "$id" ] || die "Error adding baremetal node" + id2=$(nova-baremetal-manage interface create \ + --node_id=$id --mac_address=$mac_2 --datapath_id=0 --port_no=0 \ + ) + [ $? -eq 0 ] || [ "$id2" ] || die "Error adding interface to barmetal node $id" +} + + +# Restore xtrace +$XTRACE diff --git a/lib/nova b/lib/nova index 26c5d3c603..8074153383 100644 --- a/lib/nova +++ b/lib/nova @@ -214,6 +214,11 @@ function configure_nova() { fi fi + # Prepare directories and packages for baremetal driver + if is_baremetal; then + configure_baremetal_nova_dirs + fi + if is_service_enabled quantum && is_quantum_ovs_base_plugin "$Q_PLUGIN" && ! sudo grep -q '^cgroup_device_acl' $QEMU_CONF ; then # Add /dev/net/tun to cgroup_device_acls, needed for type=ethernet interfaces cat <