Quiet down healthchecks to save battery and CPU usage, fixes ddev#1663 (

ddev#1674)

containers/ddev-dbserver/10.1/Dockerfile

@@ -1,4 +1,4 @@
-FROM mariadb:10.1.38-bionic
+FROM mariadb:10.1.40-bionic
 
 ENV MYSQL_DATABASE db
 ENV MYSQL_USER db
@@ -26,4 +26,4 @@ ENTRYPOINT ["/docker-entrypoint.sh"]
 EXPOSE 3306
 # The following line overrides any cmd entry
 CMD []
-HEALTHCHECK --interval=2s --retries=30 CMD ["/healthcheck.sh"]
+HEALTHCHECK --interval=1s --retries=30 --timeout=120s CMD ["/healthcheck.sh"]

containers/ddev-dbserver/10.1/files/docker-entrypoint.sh

@@ -4,6 +4,7 @@ set -eu
 set -o pipefail
 
 SOCKET=/var/tmp/mysql.sock
+rm -f /tmp/healthy
 
 # Wait for mysql server to be ready.
 function serverwait {

containers/ddev-dbserver/10.1/files/healthcheck.sh

@@ -3,6 +3,25 @@
 ## mysql health check for docker. original source: https://github.com/docker-library/healthcheck/blob/master/mysql/docker-healthcheck
 
 set -eo pipefail
+sleeptime=59
 
-mysql --host=127.0.0.1 -udb -pdb --database=db -e "SHOW DATABASES LIKE 'db';" >/dev/null
+# Since docker doesn't provide a lazy period for startup,
+# we track health. If the last check showed healthy
+# as determined by existence of /tmp/healthy, then
+# sleep at startup. This requires the timeout to be set
+# higher than the sleeptime used here.
+if [ -f /tmp/healthy ]; then
+    printf "container was previously healthy, so sleeping ${sleeptime} seconds before continuing healthcheck...  "
+    sleep ${sleeptime}
+fi
+
+
+if mysql --host=127.0.0.1 -udb -pdb --database=db -e "SHOW DATABASES LIKE 'db';" >/dev/null;  then
+    printf "healthy"
+    touch /tmp/healthy
+    exit 0
+fi
+
+rm -f /tmp/healthy
+exit 1
 

containers/ddev-dbserver/10.2/Dockerfile

@@ -1,4 +1,4 @@
-FROM mariadb:10.2.22
+FROM mariadb:10.2.25
 
 ENV MYSQL_DATABASE db
 ENV MYSQL_USER db
@@ -26,4 +26,4 @@ ENTRYPOINT ["/docker-entrypoint.sh"]
 EXPOSE 3306
 # The following line overrides any cmd entry
 CMD []
-HEALTHCHECK --interval=2s --retries=30 CMD ["/healthcheck.sh"]
+HEALTHCHECK --interval=1s --retries=30 --timeout=120s CMD ["/healthcheck.sh"]

containers/ddev-dbserver/10.2/files/docker-entrypoint.sh

@@ -5,6 +5,8 @@ set -o pipefail
 
 SOCKET=/var/tmp/mysql.sock
 
+rm -f /tmp/healthy
+
 # Wait for mysql server to be ready.
 function serverwait {
 	for i in {60..0};

containers/ddev-dbserver/10.2/files/healthcheck.sh

@@ -3,6 +3,25 @@
 ## mysql health check for docker. original source: https://github.com/docker-library/healthcheck/blob/master/mysql/docker-healthcheck
 
 set -eo pipefail
+sleeptime=59
 
-mysql --host=127.0.0.1 -udb -pdb --database=db -e "SHOW DATABASES LIKE 'db';" >/dev/null
+# Since docker doesn't provide a lazy period for startup,
+# we track health. If the last check showed healthy
+# as determined by existence of /tmp/healthy, then
+# sleep at startup. This requires the timeout to be set
+# higher than the sleeptime used here.
+if [ -f /tmp/healthy ]; then
+    printf "container was previously healthy, so sleeping ${sleeptime} seconds before continuing healthcheck...  "
+    sleep ${sleeptime}
+fi
+
+
+if mysql --host=127.0.0.1 -udb -pdb --database=db -e "SHOW DATABASES LIKE 'db';" >/dev/null;  then
+    printf "healthy"
+    touch /tmp/healthy
+    exit 0
+fi
+
+rm -f /tmp/healthy
+exit 1
 

containers/ddev-router/Dockerfile

@@ -1,4 +1,4 @@
-FROM nginx:1.15.12
+FROM nginx:1.17.0
 
 ENV MKCERT_VERSION=v1.3.0
 
@@ -34,4 +34,4 @@ ENTRYPOINT ["/app/docker-entrypoint.sh"]
 CMD ["forego", "start", "-r"]
 WORKDIR /app/
 
-HEALTHCHECK --interval=5s --retries=5 CMD /app/healthcheck.sh
+HEALTHCHECK --interval=1s --retries=10 --timeout=120s --start-period=10s CMD /app/healthcheck.sh

containers/ddev-router/docker-entrypoint.sh

@@ -1,6 +1,8 @@
 #!/bin/bash
 set -e
 
+rm -f /tmp/healthy
+
 # Warn if the DOCKER_HOST socket does not exist
 if [[ $DOCKER_HOST = unix://* ]]; then
 	socket_file=${DOCKER_HOST#unix://}

containers/ddev-router/healthcheck.sh

@@ -2,7 +2,39 @@
 set -eu
 set -o pipefail
 
-# Check nginx config
-nginx -t || exit 1
+sleeptime=59
+
+# Since docker doesn't provide a lazy period for startup,
+# we track health. If the last check showed healthy
+# as determined by existence of /tmp/healthy, then
+# sleep at startup. This requires the timeout to be set
+# higher than the sleeptime used here.
+if [ -f /tmp/healthy ]; then
+    printf "container was previously healthy, so sleeping ${sleeptime} seconds before continuing healthcheck...  "
+    sleep ${sleeptime}
+fi
+
+config=false
+connect=false
+
+if nginx -t ; then
+    config=true
+    printf "nginx config: OK  "
+fi
+
 # Check our healthcheck endpoint
-curl --fail --connect-timeout 2 --retry 2 http://127.0.0.1/healthcheck || (echo "ddev-router healthcheck endpoint not responding" && exit 2)
+if curl --fail --connect-timeout 2 --retry 2 http://127.0.0.1/healthcheck; then
+    connect=true
+    echo "nginx healthcheck endpoint: OK "
+else
+    echo "ddev-router healthcheck endpoint not responding "
+fi
+
+if [ ${config} = true -a ${connect} = true ]; then
+    printf "healthy"
+    touch /tmp/healthy
+    exit 0
+fi
+
+rm -f /tmp/healthy
+exit 1

containers/ddev-ssh-agent/Dockerfile

@@ -22,7 +22,7 @@
 # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 # THE SOFTWARE.
-FROM alpine:3.8
+FROM alpine:3.10
 
 # Install dependencies
 RUN apk add --no-cache \
@@ -48,7 +48,7 @@ RUN ln -s $SSH_KEY_DIR /home/.ssh
 
 RUN mkdir ${SOCKET_DIR} && mkdir ${SSH_KEY_DIR} && chmod 777 ${SOCKET_DIR} ${SSH_KEY_DIR}
 
-HEALTHCHECK --interval=2s --retries=5 CMD ["/healthcheck.sh"]
+HEALTHCHECK --interval=1s --retries=5 --timeout=120s CMD ["/healthcheck.sh"]
 
 VOLUME ${SOCKET_DIR}
 

containers/ddev-ssh-agent/files/entry.sh

@@ -21,6 +21,7 @@
 
 
 set -eo pipefail
+rm -f /tmp/healthy
 
 # Print a debug message if debug mode is on ($DEBUG is not empty)
 # @param message

containers/ddev-ssh-agent/files/healthcheck.sh

@@ -3,7 +3,23 @@
 # ddev-ssh-agent healthcheck
 
 set -eo pipefail
+sleeptime=59
 
 # Make sure that both socat and ssh-agent are running
-killall -0 socat && killall -0 ssh-agent
+# Since docker doesn't provide a lazy period for startup,
+# we track health. If the last check showed healthy
+# as determined by existence of /tmp/healthy, then
+# sleep at startup. This requires the timeout to be set
+# higher than the sleeptime used here.
+if [ -f /tmp/healthy ]; then
+    printf "container was previously healthy, so sleeping ${sleeptime} seconds before continuing healthcheck...  "
+    sleep ${sleeptime}
+fi
+if killall -0 socat ssh-agent; then
+    printf "healthy"
+    touch /tmp/healthy
+    exit 0
+fi
+rm -f /tmp/healthy
+exit 1
 

containers/ddev-webserver/Dockerfile

@@ -151,6 +151,6 @@ RUN chmod ugo+x /start.sh /healthcheck.sh
 
 
 EXPOSE 80 8025
-HEALTHCHECK --interval=3s --retries=6 CMD ["/healthcheck.sh"]
+HEALTHCHECK --interval=1s --retries=10 --timeout=120s CMD ["/healthcheck.sh"]
 
 CMD ["/start.sh"]

containers/ddev-webserver/scripts/healthcheck.sh

@@ -4,6 +4,49 @@
 
 set -eo pipefail
 
-curl --fail -s 127.0.0.1/phpstatus >/dev/null && printf "phpstatus: OK, " || (printf "phpstatus FAILED" && exit 1)
-ls /var/www/html >/dev/null && printf "/var/www/html: OK, " || (printf "/var/www/html access FAILED" && exit 2)
-curl --fail -s localhost:8025 >/dev/null && printf "mailhog: OK" || (printf "mailhog FAILED" && exit 3)
+sleeptime=59
+
+# Make sure that both phpstatus, mounted code, and mailhog
+# are working.
+# Since docker doesn't provide a lazy period for startup,
+# we track health. If the last check showed healthy
+# as determined by existence of /tmp/healthy, then
+# sleep at startup. This requires the timeout to be set
+# higher than the sleeptime used here.
+if [ -f /tmp/healthy ]; then
+    printf "container was previously healthy, so sleeping ${sleeptime} seconds before continuing healthcheck...  "
+    sleep ${sleeptime}
+fi
+
+phpstatus=false
+htmlaccess=false
+mailhog=false
+if curl --fail -s 127.0.0.1/phpstatus >/dev/null ; then
+    phpstatus=true
+    printf "phpstatus: OK "
+else
+    printf "phpstatus FAILED "
+fi
+
+if ls /var/www/html >/dev/null; then
+    htmlstatus=true
+    printf "/var/www/html: OK "
+else
+    printf "/var/www/html access FAILED"
+fi
+
+if curl --fail -s localhost:8025 >/dev/null; then
+    mailhog=true
+    printf "mailhog: OK " ;
+else
+    printf "mailhog FAILED "
+fi
+
+if ${phpstatus} = true -a ${htmlaccess} = true -a ${mailhog} = true ; then
+    touch /tmp/healthy
+    exit 0
+fi
+rm -f /tmp/healthy
+exit 1
+
+

containers/ddev-webserver/scripts/start.sh

@@ -2,6 +2,8 @@
 set -x
 set -o errexit nounset pipefail
 
+rm -f /tmp/healthy
+
 # If DDEV_PHP_VERSION isn't set, use a reasonable default
 DDEV_PHP_VERSION="${DDEV_PHP_VERSION:-$PHP_DEFAULT_VERSION}"
 

pkg/ddevapp/templates.go

@@ -40,9 +40,10 @@ services:
       - TZ={{ .Timezone }}
     command: "$DDEV_MARIADB_LOCAL_COMMAND"
     healthcheck:
-      interval: 5s
-      retries: 12
-      start_period: 60s
+      interval: 1s
+      retries: 10
+      start_period: 10s
+      timeout: 120s
   web:
     container_name: {{ .Plugin }}-${DDEV_SITENAME}-web
     {{ if .WebBuildContext }}
@@ -114,9 +115,10 @@ services:
     {{ range $hostname := .Hostnames }}- "ddev-router:{{ $hostname }}" 
     {{ end }}
     healthcheck:
-      interval: 4s
-      retries: 6
+      interval: 1s
+      retries: 10
       start_period: 10s
+      timeout: 120s
 {{ if .WebcacheEnabled }}
   bgsync:
     container_name: ddev-${DDEV_SITENAME}-bgsync
@@ -168,7 +170,7 @@ services:
       # HTTP_EXPOSE allows for ports accepting HTTP traffic to be accessible from <site>.ddev.site:<port>
       - HTTP_EXPOSE=${DDEV_PHPMYADMIN_PORT}:{{ .DBAPort }}
     healthcheck:
-      interval: 90s
+      interval: 120s
       timeout: 2s
       retries: 1
 
@@ -397,9 +399,10 @@ services:
       - ddev-global-cache:/mnt/ddev-global-cache:rw
     restart: "no"
     healthcheck:
-      interval: 6s
-      retries: 6
+      interval: 1s
+      retries: 10
       start_period: 10s
+      timeout: 120s
 
 networks:
    default:
@@ -427,8 +430,10 @@ services:
     environment:
       - SSH_AUTH_SOCK=/tmp/.ssh-agent/socket
     healthcheck:
-      interval: 2s
-      retries: 5
+      interval: 1s
+      retries: 2
+      start_period: 10s
+      timeout: 62s
 networks:
   default:
     external:

pkg/dockerutil/dockerutils_test.go

@@ -64,7 +64,6 @@ func TestMain(m *testing.M) {
 			},
 			Env: []string{"HOTDOG=superior-to-corndog", "POTATO=future-fry"},
 		},
-		//          "PortBindings": { "22/tcp": [{ "HostPort": "11022" }] },
 		HostConfig: &docker.HostConfig{
 			PortBindings: map[docker.Port][]docker.PortBinding{
 				"80/tcp": {
@@ -127,11 +126,11 @@ func TestGetContainerHealth(t *testing.T) {
 	healthDetail, err := ContainerWait(15, labels)
 	assert.NoError(err)
 
-	assert.Equal("phpstatus: OK, /var/www/html: OK, mailhog: OK", healthDetail)
+	assert.Equal("phpstatus: OK /var/www/html: OK mailhog: OK ", healthDetail)
 
 	status, healthDetail = GetContainerHealth(container)
 	assert.Equal(status, "healthy")
-	assert.Equal("phpstatus: OK, /var/www/html: OK, mailhog: OK", healthDetail)
+	assert.Equal("phpstatus: OK /var/www/html: OK mailhog: OK ", healthDetail)
 }
 
 // TestContainerWait tests the error cases for the container check wait loop.

pkg/version/version.go

@@ -45,13 +45,13 @@ var DockerComposeFileFormatVersion = "3.6"
 var WebImg = "drud/ddev-webserver"
 
 // WebTag defines the default web image tag for drud dev
-var WebTag = "20190626_apache" // Note that this can be overridden by make
+var WebTag = "20190626_healthcheck" // Note that this can be overridden by make
 
 // DBImg defines the default db image used for applications.
 var DBImg = "drud/ddev-dbserver"
 
 // BaseDBTag is the main tag, DBTag is constructed from it
-var BaseDBTag = "v1.9.0"
+var BaseDBTag = "20190626_healthcheck"
 
 // DBAImg defines the default phpmyadmin image tag used for applications.
 var DBAImg = "drud/phpmyadmin"
@@ -69,11 +69,11 @@ var BgsyncTag = "v1.9.0" // Note that this can be overridden by make
 var RouterImage = "drud/ddev-router"
 
 // RouterTag defines the tag used for the router.
-var RouterTag = "v1.9.0" // Note that this can be overridden by make
+var RouterTag = "20190626_healthcheck" // Note that this can be overridden by make
 
 var SSHAuthImage = "drud/ddev-ssh-agent"
 
-var SSHAuthTag = "v1.9.0"
+var SSHAuthTag = "20190626_healthcheck"
 
 // COMMIT is the actual committish, supplied by make
 var COMMIT = "COMMIT should be overridden"