Skip to content

armgit / macro

Public Imported from GitHub
0

Macro is a unified workspace for teams: email, chat, docs, tasks, agents, calls, and CRM — @-linked together with shared AI memory.

fix(macrodb): bound dev replica connections and add memory (#8684)

gbirmanedeafb36dc92
diff --git a/docs/development/infrastructure.md b/docs/development/infrastructure.md
index 6e595eea29..85ced8da0a 100644
--- a/docs/development/infrastructure.md
+++ b/docs/development/infrastructure.md
@@ -166,6 +166,14 @@ local resources.
- Only stacks listed in `services-config.json` deploy automatically. `gateway`,
`macrodb`, `doppler-projects`, `monitoring`, and a few others are manual.
+- Dev MacroDB's primary and replica are both configured as `db.t4g.large`
+ (8 GiB). Replica resizing requires a separate `macrodb` deployment; dev's
+ `applyImmediately` setting makes that resize interrupt replica connections.
+ Resize the replica before deploying the DSS pool change through the service
+ stack, and verify that old DSS tasks have drained before relying on the new
+ cap. Three dev tasks permit 60 replica connections after rollout, or 120
+ during a full overlap on later rollouts. The first rollout can permit 360:
+ three old tasks at 100 connections each plus three new tasks at 20 each.
- Services marked `bootstrap_pending` or `deploy_enabled: false` are skipped.
- **Migrations deploy before services** and must work with the code already
running ([database guide](database.md)).
diff --git a/infra/stacks/macrodb/Pulumi.dev.yaml b/infra/stacks/macrodb/Pulumi.dev.yaml
index c355870dc2..a398b8aacb 100644
--- a/infra/stacks/macrodb/Pulumi.dev.yaml
+++ b/infra/stacks/macrodb/Pulumi.dev.yaml
@@ -17,4 +17,4 @@ config:
macrodb:monitoring_interval: 60
macrodb:rds_monitoring_role_arn: arn:aws:iam::569036502058:role/rds-monitoring-role
macrodb:backup_retention_days: 1
- macrodb:read_replica_instance_size: db.t4g.medium
+ macrodb:read_replica_instance_size: db.t4g.large
diff --git a/services/document_storage_service/AGENTS.md b/services/document_storage_service/AGENTS.md
index 2054652e0a..0a519115b2 100644
--- a/services/document_storage_service/AGENTS.md
+++ b/services/document_storage_service/AGENTS.md
@@ -64,6 +64,11 @@ threads, pins, recents, notifications, folders, activity.
- Soup GraphQL reads a replica with multi-second lag; a read immediately after
a write through the same request needs the primary-reading loader, not the
default one — see `crates/graphql_soup`.
+- Dev's readonly pool keeps 2–20 connections per task, independently of the
+ primary pool's 25–100. Once all tasks use this cap, three dev tasks allow 60
+ sessions, or 120 while a rolling deployment overlaps all three. Old tasks
+ retain their previous cap until they drain during the initial rollout.
+ Production and local pools retain their existing primary/readonly limits.
- The GraphQL calendar feed (occurrence ranges and `calendarChanges`) reads the
primary, like the REST calendar routes: a `refresh_calendar` poke goes out as
a write commits, so a replica-backed delta misses the change it announces.
diff --git a/services/document_storage_service/src/server.rs b/services/document_storage_service/src/server.rs
index f20e82d2bf..8e8eac57af 100644
--- a/services/document_storage_service/src/server.rs
+++ b/services/document_storage_service/src/server.rs
@@ -247,14 +247,26 @@ pub async fn run() -> anyhow::Result<()> {
"initialized db connection"
);
+ let (readonly_min_connections, readonly_max_connections) = match config.environment {
+ // Once all tasks use this cap, three dev tasks allow 60 replica sessions,
+ // or 120 during a full rolling overlap.
+ // Keep excess reads queued in the service instead of exhausting replica memory.
+ Environment::Develop => (2, 20),
+ Environment::Production | Environment::Local => (min_connections, max_connections),
+ };
+
let readonly_db = match PgPoolOptions::new()
- .min_connections(min_connections)
- .max_connections(max_connections)
+ .min_connections(readonly_min_connections)
+ .max_connections(readonly_max_connections)
.connect(&config.database_url_readonly)
.await
{
Ok(pool) => {
- tracing::trace!("initialized readonly db connection");
+ tracing::trace!(
+ min_connections = readonly_min_connections,
+ max_connections = readonly_max_connections,
+ "initialized readonly db connection"
+ );
pool
}
Err(e) => {