From 6ce32f31c30bf637cba94e2cd3978d7b8b75b031 Mon Sep 17 00:00:00 2001
From: Sakala Venkata Krishna Rohit <rohitsakala@gmail.com>
Date: Wed, 16 Aug 2017 09:14:04 +0530
Subject: Add watchdog into testresults server to restart docker containers.

Unexpected crashing of docker containers were occuring. So, a cron script
is to be run in regular intervals which will check for status of the
deployments and restart docker if necessary. Docker restart is seen
as a solution to solve majority of crashes.

Suggest if there is a better approach.

Change-Id: I40d56180350844b4a0f766e1b10a9fabfa02d3be
Signed-off-by: Sakala Venkata Krishna Rohit <rohitsakala@gmail.com>
---
 testapi/tools/watchdog/docker_watch.sh | 178 +++++++++++++++++++++++++++++++++
 1 file changed, 178 insertions(+)
 create mode 100644 testapi/tools/watchdog/docker_watch.sh

(limited to 'testapi')

diff --git a/testapi/tools/watchdog/docker_watch.sh b/testapi/tools/watchdog/docker_watch.sh
new file mode 100644
index 0000000..d67e4b3
--- /dev/null
+++ b/testapi/tools/watchdog/docker_watch.sh
@@ -0,0 +1,178 @@
+#                                                               *
+#    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 script checks if deployments are working or and then
+# starts the specified containers in case one of the containers
+# crash. The only solution is restarting docker as of now.
+
+#!/bin/bash
+
+## List of modules
+modules=(testapi reporting)
+
+## Ports of the modules
+declare -A ports=( ["testapi"]="8082" ["reporting"]="8084")
+
+## Urls to check if the modules are deployed or not ?
+#declare -A urls=( ["testapi"]="http://testresults.opnfv.org/test/" \
+#    ["reporting"]="http://testresults.opnfv.org/reporting2/reporting/index.html")
+
+declare -A urls=( ["testapi"]="http://localhost:8082/" \
+    ["reporting"]="http://testresults.opnfv.org/reporting2/reporting/index.html")
+
+
+### Functions related to checking.
+
+function is_deploying() {
+    echo -e "Checking job statuses"
+    for module in "${modules[@]}"
+    do
+        if get_status $module; then
+            exit 0
+        fi
+    done
+}
+
+function get_status() {
+    xml=$(curl -m10 "https://build.opnfv.org/ci/job/${1}-automate-master/lastBuild/api/xml?depth=1")
+    building=$(grep -oPm1 "(?<=<building>)[^<]+" <<< "$xml")
+    if [[ $building == "false" ]]
+    then
+        return 1
+    else
+        return 0
+    fi
+}
+
+function get_docker_status() {
+    status=$(service docker status | sed -n 3p | cut -d ' ' -f5)
+    echo -e "Docker status: $status"
+    if [ $status = "active" ]
+    then
+        return 1
+    else
+        return 0
+    fi
+}
+
+function check_connectivity() {
+    echo "Checking $1 connection : $2"
+    cmd=`curl --head -m10 --request GET ${2} | grep '200 OK' > /dev/null`
+    rc=$?
+    if [[ $rc == 0 ]]; then
+        return 0
+    else
+        return 1
+    fi
+}
+
+function check_modules() {
+    echo -e "Checking modules"
+    failed_modules=()
+    for module in "${modules[@]}"
+    do
+        if ! check_connectivity $module "${urls[$module]}"; then
+            echo -e "$module failed"
+            failed_modules+=($module)
+        fi
+    done
+    if [ ! -z "$failed_modules" ]; then
+        echo -e "Failed Modules: $failed_modules"
+        return 1
+    else
+        echo -e "All modules working good"
+        exit 0
+    fi
+}
+
+### Functions related fixes.
+
+function restart_docker_fix() {
+    echo -e "Running restart_docker_fix"
+    service docker restart
+    start_containers_fix "${modules[@]}"
+}
+
+function docker_proxy_fix() {
+    echo -e "Running docker_proxy_fix"
+    fix_modules=("${@}")
+    for module in "${fix_modules[@]}"
+    do
+        echo -e "Kill docker proxy and restart containers"
+        pid=$(netstat -nlp | grep :${ports[$module]} | awk '{print $7}' | cut -d'/' -f1)
+        echo $pid
+        if [ ! -z "$pid" ]; then
+            kill $pid
+            start_containers_fix $module
+        fi
+    done
+}
+
+function start_containers_fix() {
+    echo "Runnning start_containers_fix"
+    start_modules=("${@}")
+    for module in "${start_modules[@]}"
+    do
+        echo -e "Starting a container $module"
+        sudo docker stop $module
+        sudo docker start $module
+        sleep 5
+        if ! check_connectivity $module "${urls[$module]}"; then
+            echo -e "Starting an old container $module_old"
+            sudo docker stop $module
+            sudo docker start $module"_old"
+            sleep 5
+        fi
+    done
+}
+
+### Main Flow
+
+echo -e
+echo -e "WatchDog Started"
+echo -e
+echo -e `date "+%Y-%m-%d %H:%M:%S.%N"`
+echo -e
+
+if ! is_deploying; then
+    echo -e "Jenkins Jobs running"
+    exit
+fi
+
+## If the problem is related to docker daemon
+
+if get_docker_status; then
+    restart_docker_fix
+    if ! check_modules; then
+        echo -e "Watchdog failed while restart_docker_fix"
+    fi
+    exit
+fi
+
+## If the problem is related to docker containers
+
+if ! check_modules; then
+    start_containers_fix "${failed_modules[@]}"
+fi
+
+## If the problem is related to docker proxy
+
+if ! check_modules; then
+    docker_proxy_fix "${failed_modules[@]}"
+fi
+
+## If nothing works out
+
+if ! check_modules; then
+    echo -e "Watchdog failed"
+fi
+
+sudo docker ps
+sudo docker images
\ No newline at end of file
-- 
cgit 1.2.3-korg