From 43661df7a25f8da4ad085ffa71e06b114f4dacc3 Mon Sep 17 00:00:00 2001
From: Sabine Maennel <5292683+sabinem@users.noreply.github.com>
Date: Wed, 7 Oct 2026 18:04:14 +0200
Subject: [PATCH] fix(just): give cluster::reseed's port-forward a real service
 name
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

`postgres_svc := "svc/{{postgres_sts}}"` was never interpolated: just
does not expand {{…}} inside a variable's value, so the port-forward
was handed the literal `svc/{{postgres_sts}}`. kubectl refuses it
after the app database has already been dropped, so reseed left dev
empty instead of seeded. Name the Service like the other resources and
let the port-forward add the `svc/` prefix, as exec adds `sts/`.

Also move the comment on restoring a backend found at 0 next to the
lines it explains.
---
 tools/just/cluster.just | 12 +++++++-----
 1 file changed, 7 insertions(+), 5 deletions(-)

diff --git a/tools/just/cluster.just b/tools/just/cluster.just
index 0ca3d289..fe082d98 100644
--- a/tools/just/cluster.just
+++ b/tools/just/cluster.just
@@ -9,6 +9,8 @@ backend_deploy := "hackagon-backend"
 backend_config := "hackagon-backend-config"
 backend_db_secret := "hackagon-backend-db"
 postgres_sts := "hackagon-postgresql"
+# The Service in front of postgres_sts; the bitnami chart gives both the same name.
+postgres_svc := "hackagon-postgresql"
 app_db := "hackagon"
 app_db_owner := "hackagon"
 
@@ -18,7 +20,6 @@ dev_domain := "hackagon-dev.dscompute.ch"
 
 backup_dir := env("HOME") + "/hackagon-backups"
 local_pg_port := "15432"
-postgres_svc := "svc/{{postgres_sts}}"
 
 # Runs inside the postgres pod: finds the superuser password the bitnami image
 # was started with (env var or mounted file) so it never leaves the pod.
@@ -88,13 +89,13 @@ wipe context:
     just -f "{{source_file()}}" backup "{{context}}"
 
     kc=(kubectl --context "{{context}}" -n "{{namespace}}")
+    # A backend found at 0 (e.g. left there by an interrupted run) comes back
+    # at 1: both recipes promise a running dev, and wipe needs it to recreate
+    # the schema.
     replicas="$("${kc[@]}" get "deploy/{{backend_deploy}}" -o jsonpath='{.spec.replicas}')"
     [ "${replicas:-0}" -gt 0 ] || replicas=1
 
     restore_backend() {
-        # A backend found at 0 (e.g. left there by an interrupted run)
-        # comes back at 1: both recipes promise a running dev,
-        # and wipe needs it to recreate the schema."
         echo "==> Scaling backend back to $replicas (it recreates the schema on startup)"
         "${kc[@]}" scale "deploy/{{backend_deploy}}" --replicas="$replicas"
     }
@@ -145,6 +146,7 @@ reseed context:
     db_password="$("${kc[@]}" get secret "{{backend_db_secret}}" -o jsonpath='{.data.password}' | base64 -d)"
     [ -n "$db_password" ] || { echo "✗ {{backend_db_secret}} has no password." >&2; exit 1; }
 
+    # As in wipe: a backend found at 0 comes back at 1.
     replicas="$("${kc[@]}" get "deploy/{{backend_deploy}}" -o jsonpath='{.spec.replicas}')"
     [ "${replicas:-0}" -gt 0 ] || replicas=1
 
@@ -161,7 +163,7 @@ reseed context:
     just -f "{{source_file()}}" _wipe-db "{{context}}"
 
     echo "==> Port-forwarding postgres to localhost:{{local_pg_port}}"
-    "${kc[@]}" port-forward "{{postgres_svc}}" "{{local_pg_port}}:5432" >/dev/null &
+    "${kc[@]}" port-forward "svc/{{postgres_svc}}" "{{local_pg_port}}:5432" >/dev/null &
     pf_pid=$!
 
     # Wait up to 10s for the tunnel to open.