Zero time restart (#2214)

* added healthcheck for traefik @devops * set a pretty tight healthcheck * change swarm polling interval to 1s * add retry middleware
ITISFoundation · Mar 19, 2021 · b9ba559 · b9ba559
1 parent ab36b42
commit b9ba559
Showing 1 changed file with 16 additions and 1 deletion.
diff --git a/services/docker-compose.yml b/services/docker-compose.yml
@@ -144,10 +144,14 @@ services:
         - traefik.http.middlewares.${SWARM_STACK_NAME_NO_HYPHEN}_sslheader.headers.customrequestheaders.X-Forwarded-Proto=http
         - traefik.enable=true
         - traefik.http.services.${SWARM_STACK_NAME}_webserver.loadbalancer.server.port=8080
+        - traefik.http.services.${SWARM_STACK_NAME}_webserver.loadbalancer.healthcheck.path=/v0/health
+        - traefik.http.services.${SWARM_STACK_NAME}_webserver.loadbalancer.healthcheck.interval=1000ms
+        - traefik.http.services.${SWARM_STACK_NAME}_webserver.loadbalancer.healthcheck.timeout=75ms
+        - traefik.http.middlewares.${SWARM_STACK_NAME}_webserver_retry.retry.attempts=3
         - traefik.http.routers.${SWARM_STACK_NAME}_webserver.rule=hostregexp(`{host:.+}`)
         - traefik.http.routers.${SWARM_STACK_NAME}_webserver.entrypoints=http
         - traefik.http.routers.${SWARM_STACK_NAME}_webserver.priority=1
-        - traefik.http.routers.${SWARM_STACK_NAME}_webserver.middlewares=${SWARM_STACK_NAME}_gzip@docker, ${SWARM_STACK_NAME_NO_HYPHEN}_sslheader@docker
+        - traefik.http.routers.${SWARM_STACK_NAME}_webserver.middlewares=${SWARM_STACK_NAME}_gzip@docker, ${SWARM_STACK_NAME_NO_HYPHEN}_sslheader@docker, ${SWARM_STACK_NAME}_webserver_retry
     networks:
       - default
       - interactive_services_subnet
@@ -324,6 +328,9 @@ services:
     command:
       - "--api=true"
       - "--api.dashboard=true"
+      - "--ping=true"
+      - "--entryPoints.ping.address=:9082"
+      - "--ping.entryPoint=ping"
       - "--log.level=WARNING"
       - "--accesslog=false"
       - "--metrics.prometheus=true"
@@ -340,6 +347,8 @@ services:
       - "--providers.docker.endpoint=unix:///var/run/docker.sock"
       - "--providers.docker.network=${SWARM_STACK_NAME}_default"
       - "--providers.docker.swarmMode=true"
+      # https://github.com/traefik/traefik/issues/7886
+      - "--providers.docker.swarmModeRefreshSeconds=1"
       - "--providers.docker.exposedByDefault=false"
       - "--providers.docker.constraints=Label(`io.simcore.zone`, `${TRAEFIK_SIMCORE_ZONE}`)"
       - "--tracing=true"
@@ -365,6 +374,12 @@ services:
     networks:
       - default
       - interactive_services_subnet
+    healthcheck:
+      test: wget --quiet --tries=1 --spider http://localhost:9082/ping || exit 1
+      interval: 3s
+      timeout: 1s
+      retries: 3
+      start_period: 1s
 
 volumes:
   input: {}