tripleo-ci/toci_gate_test.sh
Emilien Macchi b0cd45f7ed Add support for a new job testing Heat Convergence
Allow to configure a new job to test Heat Convergence on the undercloud.

Change-Id: I3795ce30e242ee9b5a491c58db9e2f49a3d11c8d
2017-01-10 16:37:50 -05:00

308 lines
14 KiB
Bash
Executable File

#!/usr/bin/env bash
set -eux
export PS4='+(${BASH_SOURCE}:${LINENO}): ${FUNCNAME[0]:+${FUNCNAME[0]}(): }'
# Mirrors
# NOTE(pabelanger): We have access to AFS mirrors, lets use them.
source /etc/nodepool/provider
source $(dirname $0)/scripts/common_vars.bash
NODEPOOL_MIRROR_HOST=${NODEPOOL_MIRROR_HOST:-mirror.$NODEPOOL_REGION.$NODEPOOL_CLOUD.openstack.org}
NODEPOOL_MIRROR_HOST=$(echo $NODEPOOL_MIRROR_HOST|tr '[:upper:]' '[:lower:]')
export CENTOS_MIRROR=http://$NODEPOOL_MIRROR_HOST/centos
export EPEL_MIRROR=http://$NODEPOOL_MIRROR_HOST/epel
if [ $NODEPOOL_CLOUD == 'tripleo-test-cloud-rh1' ]; then
source $(dirname $0)/scripts/rh2.env
# In order to save space remove the cached git repositories, at this point in
# CI the ones we are interested in have been cloned to /opt/stack/new. We
# can also remove some distro images cached on the images.
sudo rm -rf /opt/git /opt/stack/cache/files/mysql.qcow2 /opt/stack/cache/files/ubuntu-12.04-x86_64.tar.gz
fi
# Clean any cached yum metadata, it maybe stale
sudo yum clean all
# NOTE(pabelanger): Current hack to make centos-7 dib work.
# TODO(pabelanger): Why is python-requests installed from pip?
sudo rm -rf /usr/lib/python2.7/site-packages/requests
# Remove metrics from a previous run
rm -f /tmp/metric-start-times /tmp/metrics-data
# JOB_NAME used to be available from jenkins, we need to create it ourselves until
# we remove our reliance on it.
if [[ -z "${JOB_NAME-}" ]]; then
JOB_NAME=${WORKSPACE%/}
export JOB_NAME=${JOB_NAME##*/}
fi
# cd to toci directory so relative paths work
cd $(dirname $0)
# Only define $http_proxy if it is unset (use "-" instead of ":-" in the
# parameter expansion). This will allow an external script to override using a
# proxy by setting export http_proxy=""
export http_proxy=${http_proxy-"http://192.168.1.100:3128/"}
export GEARDSERVER=${TEBROKERIP-192.168.1.1}
export MIRRORSERVER=${MIRRORIP-192.168.1.101}
export CACHEUPLOAD=0
export INTROSPECT=0
export NODECOUNT=2
export PACEMAKER=0
export UNDERCLOUD_MAJOR_UPGRADE=0
# Whether or not we deploy an Overcloud
export OVERCLOUD=1
# NOTE(bnemec): At this time, the undercloud install + image build is taking from
# 1 hour to 1 hour and 15 minutes on the jobs I checked. The devstack gate timeout
# is 170 minutes, so subtracting 90 should leave us an hour and 20 minutes for
# the deploy. Hopefully that's enough, while still leaving some cushion to come
# in under the gate timeout so we can collect logs.
OVERCLOUD_DEPLOY_TIMEOUT=$((DEVSTACK_GATE_TIMEOUT-90))
export OVERCLOUD_SSH_USER=${OVERCLOUD_SSH_USER:-"jenkins"}
export OVERCLOUD_DEPLOY_ARGS=${OVERCLOUD_DEPLOY_ARGS:-""}
export OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS --libvirt-type=qemu -t $OVERCLOUD_DEPLOY_TIMEOUT"
export OVERCLOUD_UPDATE_ARGS=
export OVERCLOUD_PINGTEST_ARGS="--skip-pingtest-cleanup"
export UNDERCLOUD_SSL=0
export UNDERCLOUD_HEAT_CONVERGENCE=0
export UNDERCLOUD_IDEMPOTENT=0
export UNDERCLOUD_SANITY_CHECK=0
export TRIPLEO_SH_ARGS=
export NETISO_V4=0
export NETISO_V6=0
export RUN_PING_TEST=1
export RUN_TEMPEST_TESTS=0
export OVB=0
export UCINSTANCEID=NULL
export TOCIRUNNER="./toci_instack_ovb.sh"
export MULTINODE=0
export OVERCLOUD_ROLES=""
# Whether or not we run TripleO using OpenStack Infra nodes
export OSINFRA=0
export CONTROLLER_HOSTS=
export COMPUTE_HOSTS=
export SUBNODES_SSH_KEY=
export TEST_OVERCLOUD_DELETE=0
export OOOQ=0
export DEPLOY_OVB_EXTRA_NODE=0
if [[ $TOCI_JOBTYPE =~ scenario ]]; then
# note: we don't need PINGTEST_TEMPLATE here. See tripleo.sh. Though
# we need to export it for logs purpose.
export PINGTEST_TEMPLATE=
export MULTINODE_ENV_NAME=$TOCI_JOBTYPE
MULTINODE_ENV_PATH=/usr/share/openstack-tripleo-heat-templates/ci/environments/$MULTINODE_ENV_NAME.yaml
else
export PINGTEST_TEMPLATE=${PINGTEST_TEMPLATE:-"tenantvm_floatingip"}
export MULTINODE_ENV_NAME='multinode'
MULTINODE_ENV_PATH=$TRIPLEO_ROOT/tripleo-ci/test-environments/$MULTINODE_ENV_NAME.yaml
fi
if [[ "$TOCI_JOBTYPE" =~ "periodic" && "$TOCI_JOBTYPE" =~ "-ha" ]]; then
TEST_OVERCLOUD_DELETE=1
elif [[ "$TOCI_JOBTYPE" =~ "periodic" && "$TOCI_JOBTYPE" =~ "-nonha" ]]; then
UNDERCLOUD_IDEMPOTENT=1
fi
# TODO(trown) remove this when we get the multinode-oooq job[1] in a reasonable
# state. Right now it runs on every quickstart and quickstart-extras patch
# wasting resources.
# [1] https://review.openstack.org/#/c/416110
if [[ $TOCI_JOBTYPE =~ multinode-oooq ]]; then
exit 1
fi
# start dstat early
# TODO add it to the gate image building
sudo yum install -y dstat nmap-ncat #nc is for metrics
mkdir -p "$WORKSPACE/logs"
dstat -tcmndrylpg --top-cpu-adv --top-io-adv --nocolor | tee --append $WORKSPACE/logs/dstat.log > /dev/null &
disown
# Switch defaults based on the job name
for JOB_TYPE_PART in $(sed 's/-/ /g' <<< "${TOCI_JOBTYPE:-}") ; do
case $JOB_TYPE_PART in
updates)
if [[ "$TOCI_JOBTYPE" =~ 'ovb-updates' ]] ; then
NODECOUNT=2
# TODO(bnemec): Re-enable Ceph in the updates job
if [[ "${STABLE_RELEASE}" =~ ^mitaka$ ]] ; then
ENDPOINT_LIST_LOCATION=$TRIPLEO_ROOT/tripleo-ci/test-environments
CA_ENVIRONMENT_FILE=inject-trust-anchor-ipv6.yaml
else
ENDPOINT_LIST_LOCATION=/usr/share/openstack-tripleo-heat-templates/environments
CA_ENVIRONMENT_FILE=inject-trust-anchor-hiera-ipv6.yaml
fi
OVERCLOUD_DEPLOY_ARGS="
$OVERCLOUD_DEPLOY_ARGS
-e /usr/share/openstack-tripleo-heat-templates/environments/puppet-pacemaker.yaml
-e /usr/share/openstack-tripleo-heat-templates/environments/network-isolation-v6.yaml
-e $TRIPLEO_ROOT/tripleo-ci/test-environments/ipv6-network-templates/network-environment.yaml
-e $TRIPLEO_ROOT/tripleo-ci/test-environments/net-iso.yaml
-e $TRIPLEO_ROOT/tripleo-ci/test-environments/enable-tls-ipv6.yaml
-e $ENDPOINT_LIST_LOCATION/tls-endpoints-public-ip.yaml
-e $TRIPLEO_ROOT/tripleo-ci/test-environments/$CA_ENVIRONMENT_FILE
"
OVERCLOUD_UPDATE_ARGS="-e /usr/share/openstack-tripleo-heat-templates/overcloud-resource-registry-puppet.yaml $OVERCLOUD_DEPLOY_ARGS"
NETISO_V6=1
PACEMAKER=1
elif [[ "$TOCI_JOBTYPE" =~ 'nonha-multinode-updates' ]] ; then
OVERCLOUD_UPDATE_ARGS="-e /usr/share/openstack-tripleo-heat-templates/overcloud-resource-registry-puppet.yaml $OVERCLOUD_DEPLOY_ARGS"
fi
;;
upgrades)
if [ $TOCI_JOBTYPE == 'undercloud-upgrades' ] ; then
UNDERCLOUD_MAJOR_UPGRADE=1
export UNDERCLOUD_SANITY_CHECK=1
# We want to start by installing an Undercloud from the
# previous stable release.
if [ "$STABLE_RELEASE" = "newton" ]; then
STABLE_RELEASE=mitaka
elif [ -z $STABLE_RELEASE ]; then
STABLE_RELEASE=newton
fi
fi
;;
ha)
NODECOUNT=4
# In ci our overcloud nodes don't have access to an external netwrok
# --ntp-server is here to make the deploy command happy, the ci env
# is on virt so the clocks should be in sync without it.
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS --control-scale 3 --ntp-server 0.centos.pool.ntp.org -e /usr/share/openstack-tripleo-heat-templates/environments/puppet-pacemaker.yaml -e /usr/share/openstack-tripleo-heat-templates/environments/network-isolation.yaml -e $TRIPLEO_ROOT/tripleo-ci/test-environments/network-templates/network-environment.yaml -e $TRIPLEO_ROOT/tripleo-ci/test-environments/net-iso.yaml"
NETISO_V4=1
PACEMAKER=1
;;
nonha)
if [[ "${STABLE_RELEASE}" =~ ^(liberty|mitaka)$ ]] ; then
ENDPOINT_LIST_LOCATION=$TRIPLEO_ROOT/tripleo-ci/test-environments
CA_ENVIRONMENT_FILE=inject-trust-anchor.yaml
else
ENDPOINT_LIST_LOCATION=/usr/share/openstack-tripleo-heat-templates/environments
CA_ENVIRONMENT_FILE=inject-trust-anchor-hiera.yaml
fi
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e $TRIPLEO_ROOT/tripleo-ci/test-environments/enable-tls.yaml -e $ENDPOINT_LIST_LOCATION/tls-endpoints-public-ip.yaml -e $TRIPLEO_ROOT/tripleo-ci/test-environments/$CA_ENVIRONMENT_FILE --ceph-storage-scale 1 -e /usr/share/openstack-tripleo-heat-templates/environments/storage-environment.yaml"
INTROSPECT=1
NODECOUNT=3
UNDERCLOUD_SSL=1
;;
containers)
# TODO : remove this when the containers job is passing again
exit 1
TRIPLEO_SH_ARGS="--use-containers"
;;
ovb)
OVB=1
# The test env broker needs to know the instanceid of the this node so it can attach it to the provisioning network
UCINSTANCEID=$(http_proxy= curl http://169.254.169.254/openstack/2015-10-15/meta_data.json | python -c 'import json, sys; print json.load(sys.stdin)["uuid"]')
;;
ipv6)
NETISO_V4=0
NETISO_V6=1
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e /usr/share/openstack-tripleo-heat-templates/environments/network-isolation-v6.yaml -e $TRIPLEO_ROOT/tripleo-ci/test-environments/ipv6-network-templates/network-environment.yaml"
;;
convergence)
UNDERCLOUD_HEAT_CONVERGENCE=1
;;
multinode)
MULTINODE=1
TOCIRUNNER="./toci_instack_osinfra.sh"
OSINFRA=1
UNDERCLOUD_SSL=0
INTROSPECT=0
SUBNODES_SSH_KEY=/etc/nodepool/id_rsa
OVERCLOUD_DEPLOY_ARGS="--libvirt-type=qemu -t $OVERCLOUD_DEPLOY_TIMEOUT"
if [[ "$TOCI_JOBTYPE" =~ "3nodes" ]]; then
NODECOUNT=2
PACEMAKER=1
OVERCLOUD_ROLES="ControllerApi Controller"
export ControllerApi_hosts=$(sed -n 1,1p /etc/nodepool/sub_nodes)
export Controller_hosts=$(sed -n 2,2p /etc/nodepool/sub_nodes)
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e /usr/share/openstack-tripleo-heat-templates/environments/puppet-pacemaker.yaml -e /usr/share/openstack-tripleo-heat-templates/environments/deployed-server-environment.yaml -e $TRIPLEO_ROOT/tripleo-ci/test-environments/multinode-3nodes.yaml --compute-scale 0 --overcloud-ssh-user $OVERCLOUD_SSH_USER --validation-errors-nonfatal -r $TRIPLEO_ROOT/tripleo-ci/test-environments/multinode-3nodes-roles-data.yaml"
else
NODECOUNT=1
CONTROLLER_HOSTS=$(sed -n 1,1p /etc/nodepool/sub_nodes)
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e /usr/share/openstack-tripleo-heat-templates/environments/deployed-server-environment.yaml -e $MULTINODE_ENV_PATH --compute-scale 0 --overcloud-ssh-user $OVERCLOUD_SSH_USER --validation-errors-nonfatal"
fi
;;
undercloud)
TOCIRUNNER="./toci_instack_osinfra.sh"
NODECOUNT=0
OVERCLOUD=0
OSINFRA=1
RUN_PING_TEST=0
UNDERCLOUD_SSL=0
INTROSPECT=0
UNDERCLOUD_SSL=1
export UNDERCLOUD_SANITY_CHECK=1
;;
periodic)
export DELOREAN_REPO_URL=http://trunk.rdoproject.org/centos7/consistent
CACHEUPLOAD=1
OVERCLOUD_PINGTEST_ARGS=
;;
liberty|mitaka)
# This is handled in tripleo.sh (it always uses centos7-$STABLE_RELEASE/current)
# where $STABLE_RELEASE is derived in toci_instack.sh
unset DELOREAN_REPO_URL
;;
tempest)
export RUN_TEMPEST_TESTS=1
export RUN_PING_TEST=0
;;
oooq)
export OOOQ=1
TOCIRUNNER="./toci_instack_oooq.sh"
;;
esac
done
# Limit worker counts to avoid overloading our limited resources
if [[ "${STABLE_RELEASE}" =~ ^(liberty|mitaka)$ ]] ; then
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e $TRIPLEO_ROOT/tripleo-ci/test-environments/worker-config-mitaka-and-below.yaml"
else
OVERCLOUD_DEPLOY_ARGS="$OVERCLOUD_DEPLOY_ARGS -e $TRIPLEO_ROOT/tripleo-ci/test-environments/worker-config.yaml -e /usr/share/openstack-tripleo-heat-templates/environments/low-memory-usage.yaml"
fi
# If we're running an update job, regenerate the args to reflect the above changes
if [ -n "$OVERCLOUD_UPDATE_ARGS" ]; then
OVERCLOUD_UPDATE_ARGS="-e /usr/share/openstack-tripleo-heat-templates/overcloud-resource-registry-puppet.yaml $OVERCLOUD_DEPLOY_ARGS"
fi
TIMEOUT_SECS=$((DEVSTACK_GATE_TIMEOUT*60))
# ./testenv-client kill everything in its own process group it it hits a timeout
# run it in a separate group to avoid getting killed along with it
set -m
# install moreutils for timestamping postci.log with ts
# This comes from epel, so we need to install it before removing that repo
sudo yum install -y moreutils
# Ensure epel-release is not installed
sudo yum erase -y epel-release || :
if [ "$DEPLOY_OVB_EXTRA_NODE" = '1' ]; then
TEST_ENV_EXTRA_ARGS="--create-undercloud --ssh-key \"$(cat ~/.ssh/id_rsa.pub)\""
else
TEST_ENV_EXTRA_ARGS=""
fi
source $TRIPLEO_ROOT/tripleo-ci/scripts/metrics.bash
start_metric "tripleo.testenv.wait.seconds"
if [ -z "${TE_DATAFILE:-}" -a "$OSINFRA" = "0" ] ; then
# NOTE(pabelanger): We need gear for testenv, but this really should be
# handled by tox.
sudo pip install gear
# Kill the whole job if it doesn't get a testenv in 20 minutes as it likely will timout in zuul
( sleep 1200 ; [ ! -e /tmp/toci.started ] && sudo kill -9 $$ ) &
./testenv-client -b $GEARDSERVER:4730 -t $TIMEOUT_SECS \
--envsize $(($NODECOUNT+1)) --ucinstance $UCINSTANCEID \
$TEST_ENV_EXTRA_ARGS -- $TOCIRUNNER
else
$TOCIRUNNER
fi