feat(webapp): per-client database pool and connect timeout overrides

Adds optional per-client env overrides for the Prisma pool_timeout and connect_timeout, one pair each for the writer and read replica of the control-plane, legacy run-ops, and run-ops databases, falling back to the shared DATABASE_POOL_TIMEOUT / DATABASE_CONNECTION_TIMEOUT when unset.

This lets each database be tuned independently, e.g. a fail-fast connect timeout on one without changing the others. It also tags each client's queries with its specific datasource (control-plane, legacy-run-ops, or run-ops; writer or replica) so telemetry can attribute connection behavior per database. No behavior change until an override is set.

Author: ericallam

.server-changes/per-client-connection-timeouts.md

@@ -0,0 +1,6 @@
+---
+area: webapp
+type: improvement
+---
+
+Allow the database connection pool and connect timeouts to be tuned separately for each database's writer and read replica, falling back to the shared defaults when unset.

apps/webapp/app/db.server.ts

@@ -124,7 +124,15 @@ async function $transactionInner<R>(
 
 export { Prisma };
 
-function tagDatasource<T extends PrismaClient>(datasource: "writer" | "replica", client: T): T {
+type DatasourceLabel =
+  | "control-plane-writer"
+  | "control-plane-replica"
+  | "legacy-run-ops-writer"
+  | "legacy-run-ops-replica"
+  | "run-ops-writer"
+  | "run-ops-replica";
+
+function tagDatasource<T extends PrismaClient>(datasource: DatasourceLabel, client: T): T {
   return client.$extends({
     name: "datasource-tagger",
     query: {
@@ -142,7 +150,7 @@ function tagDatasource<T extends PrismaClient>(datasource: "writer" | "replica",
 // Same extension as tagDatasource but typed for RunOpsPrismaClient (different
 // generated package — does not extend @trigger.dev/database.PrismaClient).
 function tagDatasourceRunOps(
-  datasource: "writer" | "replica",
+  datasource: DatasourceLabel,
   client: RunOpsPrismaClient
 ): RunOpsPrismaClient {
   return client.$extends({
@@ -168,15 +176,17 @@ function captureInfraErrorsRunOps(client: RunOpsPrismaClient): RunOpsPrismaClien
 }
 
 export const prisma = singleton("prisma", () =>
-  captureInfrastructureErrors(tagDatasource("writer", getClient()))
+  captureInfrastructureErrors(tagDatasource("control-plane-writer", getClient()))
 );
 
 export const $replica: PrismaReplicaClient = singleton("replica", () => {
   const replica = getReplicaClient();
   // Brand ONLY a real replica so the run-store routing layer keeps replica reads off the primary.
   // No replica configured → fall back to the writer `prisma`, which must stay UNBRANDED.
   return replica
-    ? markReadReplicaClient(captureInfrastructureErrors(tagDatasource("replica", replica)))
+    ? markReadReplicaClient(
+        captureInfrastructureErrors(tagDatasource("control-plane-replica", replica))
+      )
     : prisma;
 });
 
@@ -296,27 +306,43 @@ const runOpsTopology: RunOpsTopology = singleton("runOpsTopology", () => {
       controlPlane: { writer: prisma, replica: $replica },
       buildNewWriter: (url, clientType) =>
         captureInfraErrorsRunOps(
-          tagDatasourceRunOps("writer", buildRunOpsWriterClient({ url, clientType }))
+          tagDatasourceRunOps("run-ops-writer", buildRunOpsWriterClient({ url, clientType }))
         ),
       // Brand the run-ops replica (only built for a real replica URL) so routed replica reads stay
       // off the primary. When no replica URL is set, selectRunOpsTopology reuses the writer here —
       // which this callback never touches, so the writer stays unbranded.
       buildNewReplica: (url, clientType) =>
         markReadReplicaClient(
           captureInfraErrorsRunOps(
-            tagDatasourceRunOps("replica", buildRunOpsReplicaClient({ url, clientType }))
+            tagDatasourceRunOps("run-ops-replica", buildRunOpsReplicaClient({ url, clientType }))
           )
         ),
       // Legacy client shares the exact control-plane wrapper stack (the legacy DB carries the full
       // control-plane schema); markReadReplicaClient only on a real replica URL, as with the NEW replica.
       buildLegacyWriter: (url, clientType) =>
         captureInfrastructureErrors(
-          tagDatasource("writer", buildWriterClient({ url, clientType }))
+          tagDatasource(
+            "legacy-run-ops-writer",
+            buildWriterClient({
+              url,
+              clientType,
+              poolTimeout: env.RUN_OPS_LEGACY_DATABASE_WRITER_POOL_TIMEOUT,
+              connectTimeout: env.RUN_OPS_LEGACY_DATABASE_WRITER_CONNECTION_TIMEOUT,
+            })
+          )
         ),
       buildLegacyReplica: (url, clientType) =>
         markReadReplicaClient(
           captureInfrastructureErrors(
-            tagDatasource("replica", buildReplicaClient({ url, clientType }))
+            tagDatasource(
+              "legacy-run-ops-replica",
+              buildReplicaClient({
+                url,
+                clientType,
+                poolTimeout: env.RUN_OPS_LEGACY_DATABASE_READ_REPLICA_POOL_TIMEOUT,
+                connectTimeout: env.RUN_OPS_LEGACY_DATABASE_READ_REPLICA_CONNECTION_TIMEOUT,
+              })
+            )
           )
         ),
     }
@@ -383,7 +409,12 @@ function getClient() {
   const url = env.CONTROL_PLANE_DATABASE_URL ?? env.DATABASE_URL;
   invariant(typeof url === "string", "neither CONTROL_PLANE_DATABASE_URL nor DATABASE_URL is set");
 
-  return buildWriterClient({ url, clientType: "writer" });
+  return buildWriterClient({
+    url,
+    clientType: "writer",
+    poolTimeout: env.DATABASE_WRITER_POOL_TIMEOUT,
+    connectTimeout: env.DATABASE_WRITER_CONNECTION_TIMEOUT,
+  });
 }
 
 // Generalized writer builder shared by the control-plane client and the run-ops
@@ -392,14 +423,18 @@ function getClient() {
 export function buildWriterClient({
   url,
   clientType,
+  poolTimeout,
+  connectTimeout,
 }: {
   url: string;
   clientType: string;
+  poolTimeout?: number;
+  connectTimeout?: number;
 }): PrismaClient {
   const databaseUrl = buildPrismaConnectionUrl(url, {
     connectionLimit: env.DATABASE_CONNECTION_LIMIT.toString(),
-    poolTimeout: env.DATABASE_POOL_TIMEOUT.toString(),
-    connectTimeout: env.DATABASE_CONNECTION_TIMEOUT.toString(),
+    poolTimeout: (poolTimeout ?? env.DATABASE_POOL_TIMEOUT).toString(),
+    connectTimeout: (connectTimeout ?? env.DATABASE_CONNECTION_TIMEOUT).toString(),
     applicationName: env.SERVICE_NAME,
   });
 
@@ -530,7 +565,12 @@ function getReplicaClient() {
     return;
   }
 
-  return buildReplicaClient({ url, clientType: "reader" });
+  return buildReplicaClient({
+    url,
+    clientType: "reader",
+    poolTimeout: env.DATABASE_READ_REPLICA_POOL_TIMEOUT,
+    connectTimeout: env.DATABASE_READ_REPLICA_CONNECTION_TIMEOUT,
+  });
 }
 
 // Generalized replica builder shared by the control-plane replica and the run-ops
@@ -539,14 +579,18 @@ function getReplicaClient() {
 export function buildReplicaClient({
   url,
   clientType,
+  poolTimeout,
+  connectTimeout,
 }: {
   url: string;
   clientType: string;
+  poolTimeout?: number;
+  connectTimeout?: number;
 }): PrismaClient {
   const replicaUrl = buildPrismaConnectionUrl(url, {
     connectionLimit: env.DATABASE_CONNECTION_LIMIT.toString(),
-    poolTimeout: env.DATABASE_POOL_TIMEOUT.toString(),
-    connectTimeout: env.DATABASE_CONNECTION_TIMEOUT.toString(),
+    poolTimeout: (poolTimeout ?? env.DATABASE_POOL_TIMEOUT).toString(),
+    connectTimeout: (connectTimeout ?? env.DATABASE_CONNECTION_TIMEOUT).toString(),
     applicationName: env.SERVICE_NAME,
   });
 
@@ -675,8 +719,10 @@ function buildRunOpsWriterClient({
 }): RunOpsPrismaClient {
   const databaseUrl = buildPrismaConnectionUrl(url, {
     connectionLimit: env.DATABASE_CONNECTION_LIMIT.toString(),
-    poolTimeout: env.DATABASE_POOL_TIMEOUT.toString(),
-    connectTimeout: env.DATABASE_CONNECTION_TIMEOUT.toString(),
+    poolTimeout: (env.RUN_OPS_DATABASE_WRITER_POOL_TIMEOUT ?? env.DATABASE_POOL_TIMEOUT).toString(),
+    connectTimeout: (
+      env.RUN_OPS_DATABASE_WRITER_CONNECTION_TIMEOUT ?? env.DATABASE_CONNECTION_TIMEOUT
+    ).toString(),
     applicationName: env.SERVICE_NAME,
   });
 
@@ -728,8 +774,12 @@ function buildRunOpsReplicaClient({
     connectionLimit: (
       env.RUN_OPS_DATABASE_READ_REPLICA_CONNECTION_LIMIT ?? env.DATABASE_CONNECTION_LIMIT
     ).toString(),
-    poolTimeout: env.DATABASE_POOL_TIMEOUT.toString(),
-    connectTimeout: env.DATABASE_CONNECTION_TIMEOUT.toString(),
+    poolTimeout: (
+      env.RUN_OPS_DATABASE_READ_REPLICA_POOL_TIMEOUT ?? env.DATABASE_POOL_TIMEOUT
+    ).toString(),
+    connectTimeout: (
+      env.RUN_OPS_DATABASE_READ_REPLICA_CONNECTION_TIMEOUT ?? env.DATABASE_CONNECTION_TIMEOUT
+    ).toString(),
     applicationName: env.SERVICE_NAME,
   });
 

apps/webapp/app/env.server.ts

@@ -120,6 +120,10 @@ const EnvironmentSchema = z
     DATABASE_CONNECTION_LIMIT: z.coerce.number().int().default(10),
     DATABASE_POOL_TIMEOUT: z.coerce.number().int().default(60),
     DATABASE_CONNECTION_TIMEOUT: z.coerce.number().int().default(20),
+    DATABASE_WRITER_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    DATABASE_WRITER_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
+    DATABASE_READ_REPLICA_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    DATABASE_READ_REPLICA_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
     // Dashboard-agent conversation store. Cloud points this at a dedicated
     // database; when unset it falls back to DATABASE_URL (OSS), where
     // the tables live in the isolated `trigger_dashboard_agent` schema.
@@ -186,6 +190,14 @@ const EnvironmentSchema = z
       .optional(),
     // Optional cap for the unpooled new run-ops read replica. Unset falls back to DATABASE_CONNECTION_LIMIT.
     RUN_OPS_DATABASE_READ_REPLICA_CONNECTION_LIMIT: z.coerce.number().int().optional(),
+    RUN_OPS_DATABASE_WRITER_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_DATABASE_WRITER_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_DATABASE_READ_REPLICA_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_DATABASE_READ_REPLICA_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_LEGACY_DATABASE_WRITER_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_LEGACY_DATABASE_WRITER_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_LEGACY_DATABASE_READ_REPLICA_POOL_TIMEOUT: z.coerce.number().int().optional(),
+    RUN_OPS_LEGACY_DATABASE_READ_REPLICA_CONNECTION_TIMEOUT: z.coerce.number().int().optional(),
     // Direct DSN for applying the full @trigger.dev/database migrations to the LEGACY run-ops DB, keeping
     // its schema current after the control plane moves off it. Direct, not pooled — migrations never run
     // over a pooler. Optional; unset -> the entrypoint's legacy migrate step is skipped.