Files
triggerdotdev--trigger.dev/docker/docker-compose.yml
T
Eric Allam 954ee5c572 fix(webapp): deliver realtime changes with current content when the read replica lags (#3910)
## Summary

When the realtime runs feed (the backend behind the `realtimeBackend`
feature flag) hydrates a change from a Postgres read replica, the read
can race the replica's apply of the very write that triggered it. The
delivered row then carries the previous change's content, and an
isolated final change (for example a last `metadata.set` before a run
goes quiet) is not corrected until the roughly 20 second backstop poll.
Measured against a replica with deliberate apply delay, every delivery
trailed exactly one change behind and a final change stranded for the
full backstop interval.

## Fix

Publishers stamp each change record with the committed row's
`updatedAt`, taken from writes they already perform, so the stamp costs
no extra queries. The router delays its wake hydrate until the replica's
measured lag has passed, anchored to that timestamp: a record that has
already spent longer than the lag in transit is hydrated immediately, so
only the racing leading edge ever waits. After hydrating, a tripwire
compares each row against its record's watermark. Still-stale rows are
withheld and retried briefly, and each detection feeds the lag estimate.
If retries run out, the rows are delivered anyway (liveness over
freshness) and follow-up re-hydrates emit the fresh version through the
normal working-set diff once the replica catches up, with the backstop
as the terminal net.

Replica lag is sampled reader-side only, and only while feeds are
active. Aurora reports live lag via `aurora_replica_status()`; vanilla
Postgres can only report "caught up or not" (mid-apply lag is not
honestly measurable from a replica), so tripwire observations floor the
estimate there. Deployments without a replica resolve to zero lag and
skip the gate entirely. Tunables live under
`REALTIME_BACKEND_NATIVE_REPLICA_LAG_*`, and
`realtime_native.stale_hydrates` plus
`realtime_native.replica_lag_estimate_ms` make replica health
observable.

Two adjacent fixes: a metadata update that writes nothing no longer
publishes a change record, and buffered parent and root metadata
operations now publish when the flusher writes them, so those changes
wake live feeds instead of waiting for the backstop.

For local testing, `docker-compose` gains an opt-in `database-replica`
service (compose profile `replica`) with a configurable
`recovery_min_apply_delay`, which reproduces replica-lag behavior
deterministically. With the gate disabled this rig reproduces the
one-change-behind delivery exactly; with it enabled, deliveries arrive
with current content at roughly the true replica lag, across write rates
faster and slower than the lag itself.
2026-06-12 07:34:50 +01:00

231 lines
7.6 KiB
YAML

# Core local-dev stack: the minimum a contributor needs to boot the webapp.
# For optional services (object store, observability, HTTP/2 proxy, chaos
# tooling, ClickHouse UI, extra electric shard) see ./docker-compose.extras.yml
# and the `pnpm run docker:full` script.
#
# Every host port is overridable via env vars from the root `.env` so multiple
# instances (worktrees, branch experiments) can run side by side. See the
# "Multiple instances" block in `.env.example` for the full set of knobs.
name: triggerdotdev-docker
volumes:
database-data:
database-data-alt:
database-replica-data:
redis-data:
minio-data:
clickhouse-data:
clickhouse-logs:
networks:
app_network:
external: false
services:
database:
container_name: ${CONTAINER_PREFIX:-}database
build:
context: .
dockerfile: Dockerfile.postgres
restart: always
volumes:
- ${DB_VOLUME:-database-data}:/var/lib/postgresql/data/
environment:
POSTGRES_USER: postgres
POSTGRES_PASSWORD: postgres
POSTGRES_DB: postgres
networks:
- app_network
ports:
- "${POSTGRES_HOST_PORT:-5432}:5432"
command:
- -c
- listen_addresses=*
- -c
- wal_level=logical
- -c
- shared_preload_libraries=pg_partman_bgw
# The webapp opens ~50 pooled connections per instance and Electric another
# ~40, so the default 100 is exhausted by one webapp + Electric alone. Raise
# it so multiple instances / load tests have headroom.
- -c
- max_connections=500
# Opt-in streaming read replica with configurable apply lag — a dial-a-lag rig for
# testing replica-race behavior (e.g. the realtime read-your-writes gate) locally.
# Start with: COMPOSE_PROFILES=replica pnpm run docker
# One-time primary prep (allows replication connections; additive, survives restarts):
# docker exec database bash -c 'grep -q "host replication" "$PGDATA/pg_hba.conf" || echo "host replication all all md5" >> "$PGDATA/pg_hba.conf"'
# docker exec database psql -U postgres -c "SELECT pg_reload_conf()"
# Then point the webapp at it: DATABASE_READ_REPLICA_URL=postgresql://postgres:postgres@localhost:5433/postgres
# Tune the lag via REPLICA_APPLY_DELAY (e.g. 150ms, 2s). Wipe database-replica-data to re-init.
database-replica:
container_name: ${CONTAINER_PREFIX:-}database-replica
profiles: ["replica"]
build:
context: .
dockerfile: Dockerfile.postgres
restart: always
depends_on:
- database
volumes:
- ${DB_REPLICA_VOLUME:-database-replica-data}:/var/lib/postgresql/data/
environment:
PGPASSWORD: postgres
REPLICA_APPLY_DELAY: ${REPLICA_APPLY_DELAY:-150ms}
networks:
- app_network
ports:
- "${POSTGRES_REPLICA_HOST_PORT:-5433}:5432"
entrypoint: ["bash", "-c"]
command:
- |
set -e
if [ ! -s "$$PGDATA/PG_VERSION" ]; then
echo "initializing streaming replica from 'database'..."
mkdir -p "$$PGDATA"
chown postgres:postgres "$$PGDATA"
chmod 0700 "$$PGDATA"
until gosu postgres pg_basebackup -h database -U postgres -D "$$PGDATA" -Fp -Xs -R; do
echo "primary not ready for replication (did you run the one-time pg_hba prep above?); retrying..."
rm -rf "$$PGDATA"/* 2>/dev/null || true
sleep 2
done
fi
# max_connections must be >= the primary's (hot-standby requirement).
exec docker-entrypoint.sh postgres -c hot_standby=on -c max_connections=500 -c "recovery_min_apply_delay=$$REPLICA_APPLY_DELAY"
redis:
container_name: ${CONTAINER_PREFIX:-}redis
image: redis:7@sha256:3e1b24a1a8f24ff926b15e5ace8c38a03e5657fb66e1fc7e5188e315aa5fa094
restart: always
volumes:
- redis-data:/data
networks:
- app_network
ports:
- "${REDIS_HOST_PORT:-6379}:6379"
# S3-compatible API for the local object store (large payloads / packet
# offload). Host :${MINIO_API_HOST_PORT:-9005} = S3 API,
# host :${MINIO_CONSOLE_HOST_PORT:-9006} = web console. The webapp only
# routes to it when the OBJECT_STORE_* env vars are set (see .env.example).
minio:
container_name: ${CONTAINER_PREFIX:-}minio
image: minio/minio:latest@sha256:14cea493d9a34af32f524e538b8346cf79f3321eff8e708c1e2960462bd8936e
restart: always
command: server /data --console-address ":9001"
environment:
MINIO_ROOT_USER: minioadmin
MINIO_ROOT_PASSWORD: minioadmin
volumes:
- minio-data:/data
ports:
- "${MINIO_API_HOST_PORT:-9005}:9000"
- "${MINIO_CONSOLE_HOST_PORT:-9006}:9001"
networks:
- app_network
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:9000/minio/health/live"]
interval: 5s
timeout: 5s
retries: 5
start_period: 5s
minio-init:
image: minio/mc:latest@sha256:a7fe349ef4bd8521fb8497f55c6042871b2ae640607cf99d9bede5e9bdf11727
depends_on:
minio:
condition: service_healthy
networks:
- app_network
entrypoint: /bin/sh
command:
- -c
- |
mc alias set local http://minio:9000 minioadmin minioadmin
mc mb -p local/packets || true
restart: "no"
electric:
container_name: ${CONTAINER_PREFIX:-}electric
image: electricsql/electric:1.2.4@sha256:20da3d0b0e74926c5623392db67fd56698b9e374c4aeb6cb5cadeb8fea171c36
restart: always
environment:
DATABASE_URL: postgresql://postgres:postgres@database:5432/postgres?sslmode=disable
ELECTRIC_INSECURE: true
ELECTRIC_ENABLE_INTEGRATION_TESTING: true
networks:
- app_network
ports:
- "${ELECTRIC_HOST_PORT:-3060}:3000"
depends_on:
- database
clickhouse:
image: clickhouse/clickhouse-server:25.6.2@sha256:97f0fe0f8729569e8c9d11069acee23abadeade4889f56ca3dc3df069f28cb85
restart: always
container_name: ${CONTAINER_PREFIX:-}clickhouse
ulimits:
nofile:
soft: 262144
hard: 262144
environment:
CLICKHOUSE_USER: default
CLICKHOUSE_PASSWORD: password
CLICKHOUSE_DEFAULT_ACCESS_MANAGEMENT: 1
ports:
- "${CLICKHOUSE_HTTP_HOST_PORT:-8123}:8123"
- "${CLICKHOUSE_TCP_HOST_PORT:-9000}:9000"
volumes:
- clickhouse-data:/var/lib/clickhouse
- clickhouse-logs:/var/log/clickhouse-server
- ./config/clickhouse-disable-system-logs.xml:/etc/clickhouse-server/config.d/disable-system-logs.xml:ro
networks:
- app_network
healthcheck:
test:
[
"CMD",
"clickhouse-client",
"--host",
"localhost",
"--port",
"9000",
"--user",
"default",
"--password",
"password",
"--query",
"SELECT 1",
]
interval: "3s"
timeout: "5s"
retries: "5"
start_period: "10s"
clickhouse_migrator:
build:
context: ../internal-packages/clickhouse
dockerfile: ./Dockerfile
depends_on:
clickhouse:
condition: service_healthy
networks:
- app_network
command: ["goose", "${GOOSE_COMMAND:-up}"]
# s2-lite: open-source S2 (https://s2.dev) for local realtime streams v2.
# The image is distroless (no shell), so a `wget` / `curl` healthcheck
# always reports unhealthy even when the API is responding. No other
# service depends on this one, so the healthcheck is omitted.
s2:
image: ghcr.io/s2-streamstore/s2:latest@sha256:d6ded5ca7dd619fa7c946f06e39a98f9c95c6883c8bb884e5eaa129f232c920c
command: ["lite", "--init-file", "/s2-spec.json"]
volumes:
- ./config/s2-spec.json:/s2-spec.json:ro
ports:
- "${S2_HOST_PORT:-4566}:80"
networks:
- app_network