fix(sqlite): concurrent migrate retries

crates/adapters/sqlite/src/migrations.rs

@@ -1,10 +1,34 @@
+use std::time::Duration;
+
 use domain::errors::DomainError;
 use sqlx::SqlitePool;
 
+/// Server and worker both migrate at startup, so two processes can race on the
+/// same database. The loser's migration fails (e.g. duplicate column) because
+/// the winner already applied it; re-running re-reads `_sqlx_migrations` and
+/// skips what is done. Only an error that persists across all retries is returned.
+const RETRY_BACKOFFS: [Duration; 5] = [
+    Duration::from_millis(100),
+    Duration::from_millis(250),
+    Duration::from_millis(500),
+    Duration::from_secs(1),
+    Duration::from_secs(1),
+];
+
 pub(crate) async fn run(pool: &SqlitePool) -> Result<(), DomainError> {
-    sqlx::migrate!("./migrations")
-        .set_ignore_missing(true)
-        .run(pool)
-        .await
-        .map_err(|e| DomainError::InfrastructureError(e.to_string()))
+    let mut migrator = sqlx::migrate!("./migrations");
+    migrator.set_ignore_missing(true);
+    let mut backoffs = RETRY_BACKOFFS.iter();
+    loop {
+        match migrator.run(pool).await {
+            Ok(()) => return Ok(()),
+            Err(e) => match backoffs.next() {
+                Some(delay) => {
+                    tracing::warn!("migration failed, retrying (concurrent migrator?): {e}");
+                    tokio::time::sleep(*delay).await;
+                }
+                None => return Err(DomainError::InfrastructureError(e.to_string())),
+            },
+        }
+    }
 }