feat: Leverage health checks for depends_on (#1057)

Expands the work on #1053 to all applicable services.

Author: BYK

CHANGELOG.md

@@ -5,6 +5,7 @@
 - feat: Support custom CA roots ([#27062](https://github.com/getsentry/sentry/pull/27062)), see the [docs](https://develop.sentry.dev/self-hosted/custom-ca-roots/) for more details.
 - fix: Fix `curl` image to version 7.77.0
 - upgrade: docker-compose version to 1.29.2
+- feat: Leverage health checks for depends_on
 
 ## 21.7.0
 

docker-compose.yml

@@ -1,6 +1,10 @@
 version: "3.4"
 x-restart-policy: &restart_policy
   restart: unless-stopped
+x-depends_on-healthy: &depends_on-healthy
+  condition: service_healthy
+x-depends_on-default: &depends_on-default
+  condition: service_started
 x-healthcheck-defaults: &healthcheck_defaults
   # Avoid setting the interval too small, as docker uses much more CPU than one would expect.
   # Related issues:
@@ -15,20 +19,34 @@ x-sentry-defaults: &sentry_defaults
   <<: *restart_policy
   image: "$SENTRY_IMAGE"
   depends_on:
-    - redis
-    - postgres
-    - memcached
-    - smtp
-    - snuba-api
-    - snuba-consumer
-    - snuba-outcomes-consumer
-    - snuba-sessions-consumer
-    - snuba-transactions-consumer
-    - snuba-subscription-consumer-events
-    - snuba-subscription-consumer-transactions
-    - snuba-replacer
-    - symbolicator
-    - kafka
+    redis:
+      <<: *depends_on-healthy
+    kafka:
+      <<: *depends_on-healthy
+    postgres:
+      <<: *depends_on-healthy
+    memcached:
+      <<: *depends_on-default
+    smtp:
+      <<: *depends_on-default
+    snuba-api:
+      <<: *depends_on-default
+    snuba-consumer:
+      <<: *depends_on-default
+    snuba-outcomes-consumer:
+      <<: *depends_on-default
+    snuba-sessions-consumer:
+      <<: *depends_on-default
+    snuba-transactions-consumer:
+      <<: *depends_on-default
+    snuba-subscription-consumer-events:
+      <<: *depends_on-default
+    snuba-subscription-consumer-transactions:
+      <<: *depends_on-default
+    snuba-replacer:
+      <<: *depends_on-default
+    symbolicator:
+      <<: *depends_on-default
   entrypoint: "/etc/sentry/entrypoint.sh"
   command: ["run", "web"]
   environment:
@@ -55,11 +73,11 @@ x-snuba-defaults: &snuba_defaults
   <<: *restart_policy
   depends_on:
     clickhouse:
-      condition: service_healthy
+      <<: *depends_on-healthy
     kafka:
-      condition: service_healthy
+      <<: *depends_on-healthy
     redis:
-      condition: service_healthy
+      <<: *depends_on-healthy
   image: "$SNUBA_IMAGE"
   environment:
     SNUBA_SETTINGS: docker
@@ -143,7 +161,8 @@ services:
   kafka:
     <<: *restart_policy
     depends_on:
-      - zookeeper
+      zookeeper:
+        <<: *depends_on-healthy
     image: "confluentinc/cp-kafka:5.5.0"
     environment:
       KAFKA_ZOOKEEPER_CONNECT: "zookeeper:2181"
@@ -184,7 +203,11 @@ services:
       # You might want to change this to a higher value (and ensure your host has enough memory)
       MAX_MEMORY_USAGE_RATIO: 0.3
     healthcheck:
-      test: ["CMD-SHELL", "wget -nv -t1 --spider 'http://localhost:9000/' || exit 1"]
+      test:
+        [
+          "CMD-SHELL",
+          "wget -nv -t1 --spider 'http://localhost:9000/' || exit 1",
+        ]
       interval: 3s
       timeout: 600s
       retries: 200
@@ -264,6 +287,14 @@ services:
       - "sentry-symbolicator:/data"
   web:
     <<: *sentry_defaults
+    healthcheck:
+      <<: *healthcheck_defaults
+      test:
+        - "CMD"
+        - "/bin/bash"
+        - '-c'
+        # Courtesy of https://unix.stackexchange.com/a/234089/108960
+        - 'exec 3<>/dev/tcp/127.0.0.1/9000 && echo -e "GET /_health/ HTTP/1.1\r\nhost: 127.0.0.1\r\n\r\n" >&3 && grep ok -s -m 1 <&3'
   cron:
     <<: *sentry_defaults
     command: run cron
@@ -318,9 +349,12 @@ services:
         source: ./geoip
         target: /geoip
     depends_on:
-      - kafka
-      - redis
-      - web
+      kafka:
+        <<: *depends_on-healthy
+      redis:
+        <<: *depends_on-healthy
+      web:
+        <<: *depends_on-healthy
 volumes:
   sentry-data:
     external: true