X-Git-Url: https://gerrit.opnfv.org/gerrit/gitweb?a=blobdiff_plain;f=extraconfig%2Ftasks%2Fyum_update.sh;h=a2a04e8e4134c94d88e5c61496aec26f2c3c260c;hb=26d7023a07c12df582145b9ff68c6bdce60fb31a;hp=4c87373e5cedc49711212a8f9f736055129ef5f3;hpb=09af40f5dbbdd335005640b538ed043dda75e77b;p=apex-tripleo-heat-templates.git

diff --git a/extraconfig/tasks/yum_update.sh b/extraconfig/tasks/yum_update.sh
index 4c87373e..a2a04e8e 100755
--- a/extraconfig/tasks/yum_update.sh
+++ b/extraconfig/tasks/yum_update.sh
@@ -38,38 +38,53 @@ if [[ -a "$timestamp_file" ]]; then
 fi
 touch "$timestamp_file"
 
-command_arguments=${command_arguments:-}
-
-list_updates=$(yum list updates)
-
-if [[ "$list_updates" == "" ]]; then
-    echo "No packages require updating"
-    exit 0
-fi
-
 pacemaker_status=""
-if hiera -c /etc/puppet/hiera.yaml service_names | grep -q pacemaker; then
+# We include word boundaries in order to not match pacemaker_remote
+if hiera -c /etc/puppet/hiera.yaml service_names | grep -q '\bpacemaker\b'; then
     pacemaker_status=$(systemctl is-active pacemaker)
 fi
 
-# Fix the redis/rabbit resource start/stop timeouts. See https://bugs.launchpad.net/tripleo/+bug/1633455
-# and https://bugs.launchpad.net/tripleo/+bug/1634851
+# (NB: when backporting this s/pacemaker_short_bootstrap_node_name/bootstrap_nodeid)
+# This runs before the yum_update so we are guaranteed to run it even in the absence
+# of packages to update (the check for -z "$update_identifier" guarantees that this
+# is run only on overcloud stack update -i)
 if [[ "$pacemaker_status" == "active" && \
-      "$(hiera -c /etc/puppet/hiera.yaml bootstrap_nodeid)" = "$(facter hostname)" ]] ; then
-    if pcs resource show rabbitmq | grep -E "start.*timeout=100"; then
-        pcs resource update rabbitmq op start timeout=200s
-    fi
-    if pcs resource show rabbitmq | grep -E "stop.*timeout=90"; then
-        pcs resource update rabbitmq op stop timeout=200s
-    fi
-    if pcs resource show redis | grep -E "start.*timeout=120"; then
-        pcs resource update redis op start timeout=200s
-    fi
-    if pcs resource show redis | grep -E "stop.*timeout=120"; then
-        pcs resource update redis op stop timeout=200s
+        "$(hiera -c /etc/puppet/hiera.yaml pacemaker_short_bootstrap_node_name | tr '[:upper:]' '[:lower:]')" == "$(facter hostname | tr '[:upper:]' '[:lower:]')" ]] ; then \
+    # OCF scripts don't cope with -eu
+    echo "Verifying if we need to fix up any IPv6 VIPs"
+    set +eu
+    fixup_wrong_ipv6_vip
+    ret=$?
+    set -eu
+    if [ $ret -ne 0 ]; then
+        echo "Fixing up IPv6 VIPs failed. Stopping here. (See https://bugs.launchpad.net/tripleo/+bug/1686357 for more info)"
+        exit 1
     fi
 fi
 
+command_arguments=${command_arguments:-}
+
+# Always ensure yum has full cache
+yum makecache || echo "Yum makecache failed. This can cause failure later on."
+
+# yum check-update exits 100 if updates are available
+set +e
+check_update=$(yum check-update 2>&1)
+check_update_exit=$?
+set -e
+
+if [[ "$check_update_exit" == "1" ]]; then
+    echo "Failed to check for package updates"
+    echo "$check_update"
+    exit 1
+elif [[ "$check_update_exit" != "100" ]]; then
+    echo "No packages require updating"
+    exit 0
+fi
+
+# special case https://bugs.launchpad.net/tripleo/+bug/1635205 +bug/1669714
+special_case_ovs_upgrade_if_needed
+
 if [[ "$pacemaker_status" == "active" ]] ; then
     echo "Pacemaker running, stopping cluster node and doing full package update"
     node_count=$(pcs status xml | grep -o "<nodes_configured.*/>" | grep -o 'number="[0-9]*"' | grep -o "[0-9]*")
@@ -81,6 +96,7 @@ if [[ "$pacemaker_status" == "active" ]] ; then
     fi
 else
     echo "Upgrading openstack-puppet-modules and its dependencies"
+    check_for_yum_lock
     yum -q -y update openstack-puppet-modules
     yum deplist openstack-puppet-modules | awk '/dependency/{print $2}' | xargs yum -q -y update
     echo "Upgrading other packages is handled by config management tooling"
@@ -90,24 +106,14 @@ fi
 
 command=${command:-update}
 full_command="yum -q -y $command $command_arguments"
-echo "Running: $full_command"
 
+echo "Running: $full_command"
+check_for_yum_lock
 result=$($full_command)
 return_code=$?
 echo "$result"
 echo "yum return code: $return_code"
 
-# Writes any changes caused by alterations to os-net-config and bounces the
-# interfaces *before* restarting the cluster.
-os-net-config -c /etc/os-net-config/config.json -v --detailed-exit-codes
-RETVAL=$?
-if [[ $RETVAL == 2 ]]; then
-    echo "os-net-config: interface configuration files updated successfully"
-elif [[ $RETVAL != 0 ]]; then
-    echo "ERROR: os-net-config configuration failed"
-    exit $RETVAL
-fi
-
 if [[ "$pacemaker_status" == "active" ]] ; then
     echo "Starting cluster node"
     pcs cluster start
@@ -124,15 +130,19 @@ if [[ "$pacemaker_status" == "active" ]] ; then
         fi
     done
 
-    tstart=$(date +%s)
-    while ! clustercheck; do
-        sleep 5
-        tnow=$(date +%s)
-        if (( tnow-tstart > galera_sync_timeout )) ; then
-            echo "ERROR galera sync timed out"
-            exit 1
-        fi
-    done
+    RETVAL=$( pcs resource show galera-master | grep wsrep_cluster_address | grep -q `crm_node -n` ; echo $? )
+
+    if [[ $RETVAL -eq 0 && -e /etc/sysconfig/clustercheck ]]; then
+        tstart=$(date +%s)
+        while ! clustercheck; do
+            sleep 5
+            tnow=$(date +%s)
+            if (( tnow-tstart > galera_sync_timeout )) ; then
+                echo "ERROR galera sync timed out"
+                exit 1
+            fi
+        done
+    fi
 
     echo "Waiting for pacemaker cluster to settle"
     if ! timeout -k 10 $cluster_settle_timeout crm_resource --wait; then
@@ -143,6 +153,7 @@ if [[ "$pacemaker_status" == "active" ]] ; then
     pcs status
 fi
 
-echo "Finished yum_update.sh on server $deploy_server_id at `date`"
+
+echo "Finished yum_update.sh on server $deploy_server_id at `date` with return code: $return_code"
 
 exit $return_code