Run migrations via a collab subcommand

Dockerfile

@@ -19,5 +19,7 @@ FROM debian:bullseye-slim as runtime
 RUN apt-get update; \
     apt-get install -y --no-install-recommends libcurl4-openssl-dev ca-certificates
 WORKDIR app
-COPY --from=builder /app/collab /app
+COPY --from=builder /app/collab /app/collab
+COPY --from=builder /app/crates/collab/migrations /app/migrations
+ENV MIGRATIONS_PATH=/app/migrations
 ENTRYPOINT ["/app/collab"]

Dockerfile.migrator

@@ -1,15 +0,0 @@
-# syntax = docker/dockerfile:1.2
-
-FROM rust:1.64-bullseye as builder
-WORKDIR app
-RUN --mount=type=cache,target=/usr/local/cargo/registry \
-    --mount=type=cache,target=./target \
-    cargo install sqlx-cli --root=/app --target-dir=/app/target --version 0.5.7
-
-FROM debian:bullseye-slim as runtime
-RUN apt-get update; \
-    apt-get install -y --no-install-recommends libssl1.1
-WORKDIR app
-COPY --from=builder /app/bin/sqlx /app
-COPY ./crates/collab/migrations /app/migrations
-ENTRYPOINT ["/app/sqlx", "migrate", "run"]

Procfile

@@ -1,2 +1,2 @@
 web: cd ../zed.dev && PORT=3000 npx vercel dev
-collab: cd crates/collab && cargo run
+collab: cd crates/collab && cargo run serve

crates/collab/Procfile

@@ -1,2 +0,0 @@
-collab: ./target/release/collab
-release: ./target/release/sqlx migrate run

crates/collab/k8s/manifest.template.yml

@@ -54,6 +54,8 @@ spec:
       containers:
         - name: collab
           image: "${ZED_IMAGE_ID}"
+          args:
+            - serve
           ports:
             - containerPort: 8080
               protocol: TCP

crates/collab/k8s/migrate.template.yml

@@ -10,6 +10,8 @@ spec:
       containers:
         - name: migrator
           image: ${ZED_IMAGE_ID}
+          args:
+            - migrate
           env:
             - name: DATABASE_URL
               valueFrom:

crates/collab/src/db.rs

@@ -6,8 +6,12 @@ use collections::HashMap;
 use futures::StreamExt;
 use serde::{Deserialize, Serialize};
 pub use sqlx::postgres::PgPoolOptions as DbOptions;
-use sqlx::{types::Uuid, FromRow, QueryBuilder};
+use sqlx::{
-use std::{cmp, ops::Range, time::Duration};
+    migrate::{Migrate as _, Migration, MigrationSource},
+    types::Uuid,
+    FromRow, QueryBuilder,
+};
+use std::{cmp, ops::Range, path::Path, time::Duration};
 use time::{OffsetDateTime, PrimitiveDateTime};
 
 #[async_trait]
@@ -173,6 +177,13 @@ pub trait Db: Send + Sync {
     fn as_fake(&self) -> Option<&FakeDb>;
 }
 
+#[cfg(any(test, debug_assertions))]
+pub const DEFAULT_MIGRATIONS_PATH: Option<&'static str> =
+    Some(concat!(env!("CARGO_MANIFEST_DIR"), "/migrations"));
+
+#[cfg(not(any(test, debug_assertions)))]
+pub const DEFAULT_MIGRATIONS_PATH: Option<&'static str> = None;
+
 pub struct PostgresDb {
     pool: sqlx::PgPool,
 }
@@ -187,6 +198,47 @@ impl PostgresDb {
         Ok(Self { pool })
     }
 
+    pub async fn migrate(
+        &self,
+        migrations_path: &Path,
+        ignore_checksum_mismatch: bool,
+    ) -> anyhow::Result<Vec<(Migration, Duration)>> {
+        let migrations = MigrationSource::resolve(migrations_path)
+            .await
+            .map_err(|err| anyhow!("failed to load migrations: {err:?}"))?;
+
+        let mut conn = self.pool.acquire().await?;
+
+        conn.ensure_migrations_table().await?;
+        let applied_migrations: HashMap<_, _> = conn
+            .list_applied_migrations()
+            .await?
+            .into_iter()
+            .map(|m| (m.version, m))
+            .collect();
+
+        let mut new_migrations = Vec::new();
+        for migration in migrations {
+            match applied_migrations.get(&migration.version) {
+                Some(applied_migration) => {
+                    if migration.checksum != applied_migration.checksum && !ignore_checksum_mismatch
+                    {
+                        Err(anyhow!(
+                            "checksum mismatch for applied migration {}",
+                            migration.description
+                        ))?;
+                    }
+                }
+                None => {
+                    let elapsed = conn.apply(&migration).await?;
+                    new_migrations.push((migration, elapsed));
+                }
+            }
+        }
+
+        Ok(new_migrations)
+    }
+
     pub fn fuzzy_like_string(string: &str) -> String {
         let mut result = String::with_capacity(string.len() * 2 + 1);
         for c in string.chars() {
@@ -1763,11 +1815,8 @@ mod test {
     use lazy_static::lazy_static;
     use parking_lot::Mutex;
     use rand::prelude::*;
-    use sqlx::{
+    use sqlx::{migrate::MigrateDatabase, Postgres};
-        migrate::{MigrateDatabase, Migrator},
+    use std::sync::Arc;
-        Postgres,
-    };
-    use std::{path::Path, sync::Arc};
     use util::post_inc;
 
     pub struct FakeDb {
@@ -2430,13 +2479,13 @@ mod test {
             let mut rng = StdRng::from_entropy();
             let name = format!("zed-test-{}", rng.gen::<u128>());
             let url = format!("postgres://postgres@localhost/{}", name);
-            let migrations_path = Path::new(concat!(env!("CARGO_MANIFEST_DIR"), "/migrations"));
             Postgres::create_database(&url)
                 .await
                 .expect("failed to create test db");
             let db = PostgresDb::new(&url, 5).await.unwrap();
-            let migrator = Migrator::new(migrations_path).await.unwrap();
+            db.migrate(Path::new(DEFAULT_MIGRATIONS_PATH.unwrap()), false)
-            migrator.run(&db.pool).await.unwrap();
+                .await
+                .unwrap();
             Self {
                 db: Some(Arc::new(db)),
                 url,

crates/collab/src/main.rs

@@ -9,12 +9,15 @@ mod db_tests;
 #[cfg(test)]
 mod integration_tests;
 
+use anyhow::anyhow;
 use axum::{routing::get, Router};
 use collab::{Error, Result};
 use db::{Db, PostgresDb};
 use serde::Deserialize;
 use std::{
+    env::args,
     net::{SocketAddr, TcpListener},
+    path::PathBuf,
     sync::Arc,
     time::Duration,
 };
@@ -34,22 +37,17 @@ pub struct Config {
     pub log_json: Option<bool>,
 }
 
+#[derive(Default, Deserialize)]
+pub struct MigrateConfig {
+    pub database_url: String,
+    pub migrations_path: Option<PathBuf>,
+}
+
 pub struct AppState {
     db: Arc<dyn Db>,
     config: Config,
 }
 
-impl AppState {
-    async fn new(config: Config) -> Result<Arc<Self>> {
-        let db = PostgresDb::new(&config.database_url, 5).await?;
-        let this = Self {
-            db: Arc::new(db),
-            config,
-        };
-        Ok(Arc::new(this))
-    }
-}
-
 #[tokio::main]
 async fn main() -> Result<()> {
     if let Err(error) = env::load_dotenv() {
@@ -59,24 +57,59 @@ async fn main() -> Result<()> {
         );
     }
 
-    let config = envy::from_env::<Config>().expect("error loading config");
+    match args().skip(1).next().as_deref() {
-    init_tracing(&config);
+        Some("version") => {
-    let state = AppState::new(config).await?;
+            println!("collab v{VERSION}");
+        }
+        Some("migrate") => {
+            let config = envy::from_env::<MigrateConfig>().expect("error loading config");
+            let db = PostgresDb::new(&config.database_url, 5).await?;
 
-    let listener = TcpListener::bind(&format!("0.0.0.0:{}", state.config.http_port))
+            let migrations_path = config
-        .expect("failed to bind TCP listener");
+                .migrations_path
-    let rpc_server = rpc::Server::new(state.clone(), None);
+                .as_deref()
+                .or(db::DEFAULT_MIGRATIONS_PATH.map(|s| s.as_ref()))
+                .ok_or_else(|| anyhow!("missing MIGRATIONS_PATH environment variable"))?;
 
-    rpc_server.start_recording_project_activity(Duration::from_secs(5 * 60), rpc::RealExecutor);
+            let migrations = db.migrate(&migrations_path, false).await?;
+            for (migration, duration) in migrations {
+                println!(
+                    "Ran {} {} {:?}",
+                    migration.version, migration.description, duration
+                );
+            }
 
-    let app = api::routes(&rpc_server, state.clone())
+            return Ok(());
-        .merge(rpc::routes(rpc_server))
+        }
-        .merge(Router::new().route("/", get(handle_root)));
+        Some("serve") => {
+            let config = envy::from_env::<Config>().expect("error loading config");
+            let db = PostgresDb::new(&config.database_url, 5).await?;
 
-    axum::Server::from_tcp(listener)?
+            init_tracing(&config);
-        .serve(app.into_make_service_with_connect_info::<SocketAddr>())
+            let state = Arc::new(AppState {
-        .await?;
+                db: Arc::new(db),
+                config,
+            });
 
+            let listener = TcpListener::bind(&format!("0.0.0.0:{}", state.config.http_port))
+                .expect("failed to bind TCP listener");
+
+            let rpc_server = rpc::Server::new(state.clone(), None);
+            rpc_server
+                .start_recording_project_activity(Duration::from_secs(5 * 60), rpc::RealExecutor);
+
+            let app = api::routes(&rpc_server, state.clone())
+                .merge(rpc::routes(rpc_server))
+                .merge(Router::new().route("/", get(handle_root)));
+
+            axum::Server::from_tcp(listener)?
+                .serve(app.into_make_service_with_connect_info::<SocketAddr>())
+                .await?;
+        }
+        _ => {
+            Err(anyhow!("usage: collab <version | migrate | serve>"))?;
+        }
+    }
     Ok(())
 }
 

script/bootstrap

@@ -7,7 +7,7 @@ echo "creating database..."
 script/sqlx database create
 
 echo "migrating database..."
-script/sqlx migrate run
+cargo run -p collab -- migrate
 
 echo "seeding database..."
 script/seed-db

script/deploy

@@ -1,12 +1,7 @@
 #!/bin/bash
 
-# Prerequisites:
-#
-# - Log in to the DigitalOcean API, either interactively, by running
-#   `doctl auth init`, or by setting the `DIGITALOCEAN_ACCESS_TOKEN`
-#   environment variable.
-
 set -eu
+source script/lib/deploy-helpers.sh
 
 if [[ $# < 2 ]]; then
   echo "Usage: $0 <production|staging|preview> <tag-name>"
@@ -15,28 +10,8 @@ fi
 export ZED_KUBE_NAMESPACE=$1
 COLLAB_VERSION=$2
 
-ENV_FILE="crates/collab/k8s/environments/${ZED_KUBE_NAMESPACE}.sh"
+export_vars_for_environment $ZED_KUBE_NAMESPACE
-if [[ ! -f $ENV_FILE ]]; then
+export ZED_IMAGE_ID=$(image_id_for_version $COLLAB_VERSION)
-  echo "Invalid environment name '${ZED_KUBE_NAMESPACE}'"
+target_zed_kube_cluster
-  exit 1
-fi
-export $(cat $ENV_FILE)
-
-if [[ ! $COLLAB_VERSION =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
-  echo "Invalid version number '$COLLAB_VERSION'"
-  exit 1
-fi
-TAG_NAMES=$(doctl registry repository list-tags collab --no-header --format Tag)
-if ! $(echo "${TAG_NAMES}" | grep -Fqx v${COLLAB_VERSION}); then
-  echo "No such image tag: 'zed/collab:v${COLLAB_VERSION}'"
-  echo "Found tags"
-  echo "${TAG_NAMES}"
-  exit 1
-fi
-export ZED_IMAGE_ID="registry.digitalocean.com/zed/collab:v${COLLAB_VERSION}"
-
-if [[ $(kubectl config current-context 2> /dev/null) != do-nyc1-zed-1 ]]; then
-  doctl kubernetes cluster kubeconfig save zed-1
-fi
 
 envsubst < crates/collab/k8s/manifest.template.yml | kubectl apply -f -

script/deploy-migration

@@ -1,42 +1,20 @@
 #!/bin/bash
 
-# Prerequisites:
-#
-# - Log in to the DigitalOcean docker registry
-#   doctl registry login
-#
-# - Target the `zed-1` kubernetes cluster
-#   doctl kubernetes cluster kubeconfig save zed-1
-
 set -eu
+source script/lib/deploy-helpers.sh
 
-if [[ $# < 1 ]]; then
+if [[ $# < 2 ]]; then
-  echo "Usage: $0 [production|staging|...]"
+  echo "Usage: $0 <production|staging|preview> <tag-name>"
   exit 1
 fi
-
 export ZED_KUBE_NAMESPACE=$1
-ENV_FILE="crates/collab/k8s/environments/${ZED_KUBE_NAMESPACE}.sh"
+COLLAB_VERSION=$2
-if [[ ! -f $ENV_FILE ]]; then
-  echo "Invalid environment name '${ZED_KUBE_NAMESPACE}'"
-  exit 1
-fi
 
-if [[ -n $(git status --short) ]]; then
+export_vars_for_environment $ZED_KUBE_NAMESPACE
-  echo "Cannot deploy with uncommited changes"
+export ZED_IMAGE_ID=$(image_id_for_version ${COLLAB_VERSION})
-  exit 1
+export ZED_MIGRATE_JOB_NAME=zed-migrate-${COLLAB_VERSION}
-fi
+target_zed_kube_cluster
-
-git_sha=$(git rev-parse HEAD)
-export ZED_IMAGE_ID=registry.digitalocean.com/zed/zed-migrator:${ZED_KUBE_NAMESPACE}-${git_sha}
-export ZED_MIGRATE_JOB_NAME=zed-migrate-${git_sha}
-
-docker build . \
-  --file ./Dockerfile.migrator \
-  --tag $ZED_IMAGE_ID
-docker push $ZED_IMAGE_ID
 
 envsubst < crates/collab/k8s/migrate.template.yml | kubectl apply -f -
-
 pod=$(kubectl --namespace=${ZED_KUBE_NAMESPACE} get pods --selector=job-name=${ZED_MIGRATE_JOB_NAME} --output=jsonpath='{.items[*].metadata.name}')
 echo "pod:" $pod

script/lib/deploy-helpers.sh

@@ -0,0 +1,43 @@
+# Prerequisites:
+#
+# - Log in to the DigitalOcean API, either interactively, by running
+#   `doctl auth init`, or by setting the `DIGITALOCEAN_ACCESS_TOKEN`
+#   environment variable.
+
+function export_vars_for_environment {
+  local environment=$1
+  local env_file="crates/collab/k8s/environments/${environment}.sh"
+  if [[ ! -f $env_file ]]; then
+    echo "Invalid environment name '${environment}'"
+    exit 1
+  fi
+  export $(cat $env_file)
+}
+
+function image_id_for_version {
+  local version=$1
+  if [[ ! ${version} =~ ^[0-9]+\.[0-9]+\.[0-9]+$ ]]; then
+    echo "Invalid version number '${version}'"
+    exit 1
+  fi
+  TAG_NAMES=$(doctl registry repository list-tags collab --no-header --format Tag)
+  if ! $(echo "${TAG_NAMES}" | grep -Fqx v${version}); then
+    echo "No such image tag: 'zed/collab:v${version}'"
+    echo "Found tags"
+    echo "${TAG_NAMES}"
+    exit 1
+  fi
+  
+  echo "registry.digitalocean.com/zed/collab:v${version}"
+}
+
+function version_for_image_id {
+  local image_id=$1
+  echo $image_id | cut -d: -f2
+}
+
+function target_zed_kube_cluster {
+  if [[ $(kubectl config current-context 2> /dev/null) != do-nyc1-zed-1 ]]; then
+    doctl kubernetes cluster kubeconfig save zed-1
+  fi
+}