-#!/bin/bash
-function wait_for() {
- local cmdstr=$@
- local total_attempts=360
- local sleep_time=10
- local attempt=1
- echo "[NOTE] Waiting for cmd to return success: ${cmdstr}"
- while ((attempt <= total_attempts)); do
- eval "${cmdstr}" && break || true
- echo -n '.'; sleep "${sleep_time}"
- ((attempt+=1))
+#!/bin/bash -e
+##############################################################################
+# Copyright (c) 2017 Mirantis Inc., Enea AB and others.
+# All rights reserved. This program and the accompanying materials
+# are made available under the terms of the Apache License, Version 2.0
+# which accompanies this distribution, and is available at
+# http://www.apache.org/licenses/LICENSE-2.0
+##############################################################################
+
+CI_DEBUG=${CI_DEBUG:-0}; [[ "${CI_DEBUG}" =~ (false|0) ]] || set -x
+ERASE_ENV=${ERASE_ENV:-0}
+
+# shellcheck disable=SC1090
+source "$(dirname "${BASH_SOURCE[0]}")/../../scripts/lib.sh"
+
+# Wait for MaaS commissioning/deploy to finish, retry on failure
+function maas_fixup() {
+ local statuscmd="salt 'mas01*' --out yaml state.apply maas.machines.status"
+ # shellcheck disable=SC2155
+ local ncount=$(salt --out yaml 'mas01*' pillar.get maas:region:machines | \
+ grep -cE '^\s{2}\w+:$')
+ wait_for 180 "${statuscmd} | tee /dev/stderr | " \
+ "grep -Eq '((Deployed|Ready): ${ncount}|status: (Failed|Allocated))'"
+ # shellcheck disable=SC2155
+ local statusout=$(eval "${statuscmd}")
+
+ # shellcheck disable=SC2155
+ local fcnodes=$(echo "${statusout}" | \
+ grep -Pzo 'status: Failed commissioning\n\s+system_id: \K.+\n')
+ for node_system_id in ${fcnodes}; do
+ salt -C 'mas01*' state.apply maas.machines.delete \
+ pillar="{'system_id': '${node_system_id}'}"
+ sleep 30
done
+ if [ -n "${fcnodes}" ]; then
+ salt -C 'mas01*' state.apply maas.machines
+ return 1
+ fi
+
+ # shellcheck disable=SC2155
+ local fdnodes=$(echo "${statusout}" | \
+ grep -Pzo 'status: (Failed deployment|Allocated)\n\s+system_id: \K.+\n')
+ for node_system_id in ${fdnodes}; do
+ salt -C 'mas01*' state.apply maas.machines.mark_broken_fixed \
+ pillar="{'system_id': '${node_system_id}'}"
+ sleep 30
+ done
+ if [ -n "${fdnodes}" ]; then
+ salt -C 'mas01*' state.apply maas.machines.deploy
+ return 1
+ fi
+
+ return 0
}
+# Optionally destroy MaaS machines from a previous run
+if [ "${ERASE_ENV}" -gt 1 ]; then
+ dnodes=$(salt 'mas01*' --out yaml state.apply maas.machines.status | \
+ grep -Pzo '\s+system_id: \K.+\n')
+ for node_system_id in ${dnodes}; do
+ salt -C 'mas01*' state.apply maas.machines.delete \
+ pillar="{'system_id': '${node_system_id}'}"
+ sleep 30
+ done
+fi
+
# MaaS rack/region controller, node commissioning
salt -C 'mas01*' cmd.run "add-apt-repository ppa:maas/stable"
salt -C 'mas01*' state.apply linux.network.interface
salt -C 'mas01*' state.apply maas.pxe_nat
salt -C 'mas01*' state.apply maas.cluster
-salt -C 'cfg01*' cmd.run \
- "route add -net 192.168.11.0/24 gw ${MAAS_IP:-192.168.10.3}"
+salt -C 'cfg01*' state.apply maas.pxe_route
-wait_for "salt -C 'mas01*' state.apply maas.region"
+wait_for 10 "salt -C 'mas01*' state.apply maas.region"
salt -C 'mas01*' state.apply maas.machines
-# TODO: relax cond, as this is not re-entrant (e.g. nodes already deployed)
-wait_for "salt 'mas01*' --out yaml state.apply maas.machines.status | " \
- "tee /dev/stderr | fgrep -q 'Ready: 5'"
+wait_for 10 maas_fixup
+
+# cleanup outdated salt keys
+salt-key --out yaml | awk '!/^(minions|- cfg01|- mas01)/ {print $2}' | \
+ xargs -I{} salt-key -yd {}
# MaaS node deployment
salt -C 'mas01*' state.apply maas.machines.deploy
-wait_for "salt 'mas01*' --out yaml state.apply maas.machines.status | " \
- "tee /dev/stderr | fgrep -q 'Deployed: 5'"
+wait_for 10 maas_fixup
salt -C 'mas01*' pillar.item\
maas:region:admin:username \
maas:region:admin:password
-# KVM, compute node prereqs (libvirt first), VCP deployment
-salt -C '* and not cfg01* and not mas01*' saltutil.sync_all
-
-salt -C 'kvm*' pkg.install bridge-utils
-salt -C 'kvm*' state.apply linux.network
-salt -C 'kvm*' system.reboot
-wait_for "! salt '*' test.ping | tee /dev/stderr | fgrep -q 'Not connected'"
-
-salt -C '* and not cfg01* and not mas01*' state.apply linux,ntp
-
-salt -C 'kvm*' state.sls libvirt
-
-salt -C '* and not cfg01* and not mas01*' state.apply salt
-salt -C 'kvm*' state.sls salt.control
-
-vcp_nodes=$(salt --out yaml 'kvm01*' pillar.get salt:control:cluster:internal:node | awk '/\s+\w+:$/ {gsub(/:$/, "*"); print $1}')
-
-# Check all vcp nodes are available
+# Check all baremetal nodes are available
rc=1
-while [ $rc -ne 0 ]; do
+attempt=0
+total_attempts=10
+while [ $rc -ne 0 ] && [ ${attempt} -lt ${total_attempts} ]; do
+ bm_nodes=$(salt --out yaml 'mas01*' pillar.get maas:region:machines | \
+ awk '/^\s+\w+[[:digit:]]+:$/ {gsub(/:$/, "*"); print $1}')
rc=0
- for node in $vcp_nodes; do
- salt "$node" test.ping 2>/dev/null 1>&2 || { rc=$?; break; };
+ for node in $bm_nodes; do
+ salt "$node" test.ping 2>/dev/null || { rc=$?; break; };
done
+ sleep 5
+ ((attempt+=1))
done
-salt -C '* and not cfg01* and not mas01*' saltutil.sync_all
-salt -C '* and not cfg01* and not mas01*' state.apply salt
-salt -C '* and not cfg01* and not mas01*' state.apply linux,ntp
+wait_for 10 "salt -C '* and not cfg01* and not mas01*' saltutil.sync_all"