Ignore default db for migration/management

doc/postgres/migration.md

@@ -0,0 +1,287 @@
+# Migrating PostgreSQL cluster to a new major version
+
+## Summary
+
+1. Dump from a replica
+2. Restore to fresh VM running new major version
+3. Add logical replication for delta sync from current/old primary
+4. Switch primary to new server
+5. Remove logical replication on new server
+
+## Runbook
+
+* Primary host: `PRIMARY_HOST`
+* Replica host: `REPLICA_HOST`
+* New PG14 host: `NEW_HOST`
+* PostgreSQL superuser: `postgres`
+* Running locally on each machine via `sudo -u postgres`
+
+Adjust hostnames/IPs/etc. where needed.
+
+---
+
+### 🟢 0. PRIMARY — Pre-checks
+
+```bash
+sudo -u postgres psql -c "SHOW wal_level;"
+sudo -u postgres psql -c "SHOW max_replication_slots;"
+```
+
+If needed, edit config:
+
+```bash
+sudo -u postgres vi $PGDATA/postgresql.conf
+```
+
+Ensure:
+
+```conf
+wal_level = logical
+max_replication_slots = 10
+```
+
+Restart if changed:
+
+```bash
+sudo systemctl restart postgresql
+```
+
+---
+
+### 🔵🟡 3. Create keypair for syncing dump later
+
+🔵 On NEW_HOST:
+
+```bash
+sudo mkdir -p /home/postgres/.ssh && \
+sudo chown -R postgres:postgres /home/postgres && \
+sudo chmod 700 /home/postgres/.ssh && \
+sudo -u postgres bash -c 'ssh-keygen -t ecdsa -b 256 -f /home/postgres/.ssh/id_ecdsa -N "" -C "postgres@$(hostname)"' && \
+sudo cat /home/postgres/.ssh/id_ecdsa.pub
+```
+
+Copy the public key from the above output
+
+🟡 On replica:
+
+```bash
+sudo mkdir -p /home/postgres/.ssh && \
+sudo chown -R postgres:postgres /home/postgres && \
+sudo chmod 700 /home/postgres/.ssh && \
+echo [public_key] | sudo tee /home/postgres/.ssh/authorized_keys > /dev/null && \
+sudo chmod 700 /home/postgres/.ssh
+```
+
+---
+
+### 🟢 1. PRIMARY — Create publication and replication slots
+
+```bash
+sudo -u postgres pg_create_replication_publications
+```
+
+or
+
+```bash
+sudo -u postgres pg_create_replication_publication [db_name]
+```
+
+Listing publications and slots:
+
+```bash
+sudo -u postgres pg_list_replication_publications
+sudo -u postgres pg_list_replication_slots
+```
+
+---
+
+### 🟡 3. REPLICA — Pause replication
+
+```bash
+sudo -u postgres psql -c "SELECT pg_wal_replay_pause();"
+```
+
+Verify:
+
+```bash
+sudo -u postgres psql -c "SELECT pg_is_wal_replay_paused();"
+```
+
+---
+
+### 🟡 4. REPLICA — Run dump
+
+```bash
+sudo -u postgres pg_dump_all_databases
+```
+
+or
+
+```bash
+sudo -u postgres bash -c "pg_dumpall --globals-only > /tmp/globals.sql"
+sudo -u postgres pg_dump_database [db_name]
+```
+
+---
+
+### 🟡 5. REPLICA — Resume replication
+
+```bash
+sudo -u postgres psql -c "SELECT pg_wal_replay_resume();"
+```
+
+---
+
+### 🔵 6. COPY dumps to NEW HOST
+
+From NEW_HOST:
+
+```bash
+export REPLICA_HOST=[private_ip] && \
+cd /tmp && \
+sudo -u postgres scp "postgres@$REPLICA_HOST:/tmp/globals.sql" . && \
+sudo -u postgres scp "postgres@$REPLICA_HOST:/tmp/dump_*.tar.zst" .
+```
+
+---
+
+### 🔵 7. NEW HOST (PostgreSQL 14) — Restore
+
+#### 7.1 Restore globals
+
+```bash
+sudo -u postgres psql -f /tmp/globals.sql
+```
+
+---
+
+#### 7.2 Create databases
+
+```bash
+sudo -u postgres psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1')" | \
+xargs -I{} sudo -u postgres createdb {}
+```
+
+or
+
+```bash
+sudo -u postgres createdb [db_name]
+```
+
+---
+
+#### 7.3 Restore each database
+
+```bash
+sudo -u postgres pg_restore_all_databases
+```
+
+or
+
+```bash
+sudo -u postgres pg_restore_database [db_name]
+```
+
+---
+
+### 🔵 8. NEW HOST — Create subscriptions
+
+```bash
+sudo -u postgres pg_create_replication_subscriptions
+```
+
+or
+
+```bash
+sudo -u postgres pg_create_replication_subscription [db_name]
+```
+
+---
+
+### 🔵 9. NEW HOST — Monitor replication
+
+```bash
+sudo -u postgres pg_list_replication_subscriptions
+```
+
+---
+
+### 🔴 11. CUTOVER
+
+#### 11.1 Stop writes on old primary
+
+Put app(s) in maintenance mode, stop the app/daemons.
+
+---
+
+#### 11.2 Wait for replication to catch up
+
+TODO: not the best way to check, since WAL LSNs keep increasing
+
+```bash
+sudo -u postgres psql -d [db_name] -c "SELECT * FROM pg_stat_subscription;"
+```
+
+---
+
+#### 11.3 Fix sequences
+
+Run per DB:
+
+```bash
+sudo -u postgres pg_fix_sequences_in_all_databases
+```
+
+or
+
+```bash
+sudo -u postgres pg_fix_sequences [db_name]
+```
+
+---
+
+#### 11.4 Point app to NEW_HOST
+
+1. Update `pg.kosmos.local` in `/etc/hosts` on app server(s). For example:
+
+   ```bash
+   export NEW_PG_PRIMARY=[private_ip]
+   bundle exec knife ssh roles:ejabberd -a knife_zero.host "sudo sed -r \"s/^[0-9]+\.[0-9]+\.[0-9]+\.[0-9]+\s(pg.kosmos.local)/$NEW_PG_PRIMARY\t\1/\" -i /etc/hosts"
+   ```
+
+   Or override node attribute(s) if necessary and/or approporiate.
+
+2. Start the app/daemons, and deactivate maintenance mode.
+
+---
+
+### 🧹 12. CLEANUP NEW_HOST
+
+```bash
+sudo -u postgres pg_drop_replication_subscriptions
+```
+
+---
+
+### 🧹 13. CLEANUP PRIMARY
+
+TODO: Looks like slots are dropped automatically, when subscriptions are dropped
+
+```bash
+sudo -u postgres pg_drop_replication_publications
+```
+
+---
+
+### 🧹 13. CLEANUP Chef
+
+Once all apps/databases are migrated, update the role in the node
+config of the new primary to 'postgres_primary' and converge it.
+
+Also delete the old primary node config from the Chef repo.
+
+---
+
+### ✅ DONE
+
+---

roles/postgresql_replica_logical.rb

@@ -0,0 +1,8 @@
+name "postgresql_replica_logical"
+
+run_list [
+  "kosmos_postgresql::hostsfile",
+  "kosmos_postgresql::replica_logical",
+  "kosmos_postgresql::firewall",
+  "kosmos_postgresql::management_scripts"
+]

site-cookbooks/kosmos_drone/attributes/default.rb

@@ -1,2 +1,6 @@
 node.default["kosmos_drone"]["domain"] = "drone.kosmos.org"
 node.default["kosmos_drone"]["upstream_port"] = 80
+node.default["kosmos_drone"]["pg_host"] = "pg.kosmos.local"
+node.default["kosmos_drone"]["pg_port"] = 5432
+node.default["kosmos_drone"]["pg_db"] = "drone"
+node.default["kosmos_drone"]["pg_user"] = "drone"

site-cookbooks/kosmos_drone/recipes/default.rb

@@ -9,11 +9,11 @@ credentials = data_bag_item("credentials", "drone")
 drone_credentials = data_bag_item('credentials', 'drone')
 
 postgres_config = {
-  username: "drone",
-  password: drone_credentials["postgresql_password"],
-  host: "pg.kosmos.local",
-  port: 5432,
-  database: "drone"
+  host: node["kosmos_drone"]["pg_host"],
+  port: node["kosmos_drone"]["pg_port"],
+  database: node["kosmos_drone"]["pg_db"],
+  username: node["kosmos_drone"]["pg_user"],
+  password: drone_credentials["postgresql_password"]
 }
 
 directory deploy_path do

site-cookbooks/kosmos_postgresql/attributes/default.rb

@@ -1,3 +1,8 @@
+node.default['kosmos_postgresql']['postgresql_version'] = "14"
+
 # This is set to false by default, and set to true in the server resource
 # for replicas.
 node.default['kosmos_postgresql']['ready_to_set_up_replica'] = false
+
+# Address space from which clients are allowed to connect
+node.default['kosmos_postgresql']['access_addr'] = "10.1.1.0/24"

site-cookbooks/kosmos_postgresql/files/create_publication.sh

@@ -0,0 +1,31 @@
+#!/bin/bash
+set -euo pipefail
+
+DB_NAME="${1:?Usage: $0 <database_name>}"
+
+echo "== Processing DB: $DB_NAME =="
+
+# Create publication (idempotent)
+psql -d "$DB_NAME" -v ON_ERROR_STOP=1 <<'SQL'
+DO $$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_publication WHERE pubname = 'migrate_pub'
+  ) THEN
+    CREATE PUBLICATION migrate_pub FOR ALL TABLES;
+  END IF;
+END
+$$;
+SQL
+
+# Create logical replication slot (idempotent-ish)
+SLOT="migrate_slot_${DB_NAME}"
+
+if ! psql -d "$DB_NAME" -Atqc "SELECT 1 FROM pg_replication_slots WHERE slot_name = '$SLOT'" | grep -q 1; then
+  echo "  Creating slot: $SLOT"
+  psql -d "$DB_NAME" -c "SELECT pg_create_logical_replication_slot('$SLOT', 'pgoutput');"
+else
+  echo "  Slot already exists: $SLOT"
+fi
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/files/create_publications.sh

@@ -0,0 +1,34 @@
+#!/bin/bash
+set -e
+
+echo "== Creating publication in each database =="
+
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "Processing DB: $db"
+
+  # Create publication (idempotent)
+  psql -d "$db" -v ON_ERROR_STOP=1 <<SQL
+DO \$\$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_publication WHERE pubname = 'migrate_pub'
+  ) THEN
+    CREATE PUBLICATION migrate_pub FOR ALL TABLES;
+  END IF;
+END
+\$\$;
+SQL
+
+  # Create logical replication slot (idempotent-ish)
+  SLOT="migrate_slot_${db}"
+
+  if ! psql -d "$db" -Atqc "SELECT 1 FROM pg_replication_slots WHERE slot_name = '$SLOT'" | grep -q 1; then
+    echo "  Creating slot: $SLOT"
+    psql -d "$db" -c "SELECT pg_create_logical_replication_slot('$SLOT', 'pgoutput');"
+  else
+    echo "  Slot already exists: $SLOT"
+  fi
+
+done
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/files/drop_publications.sh

@@ -0,0 +1,34 @@
+#!/bin/bash
+set -e
+
+echo "== Dropping subscriptions slots and publications =="
+
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "Processing DB: $db"
+
+  SLOT="migrate_slot_${db}"
+
+  # Drop slot if exists
+  if psql -d "$db" -Atqc "SELECT 1 FROM pg_replication_slots WHERE slot_name = '$SLOT'" | grep -q 1; then
+    echo "  Dropping slot: $SLOT"
+    psql -d "$db" -c "SELECT pg_drop_replication_slot('$SLOT');"
+  else
+    echo "  Slot not found: $SLOT"
+  fi
+
+  # Drop publication if exists
+  psql -d "$db" -v ON_ERROR_STOP=1 <<SQL
+DO \$\$
+BEGIN
+  IF EXISTS (
+    SELECT 1 FROM pg_publication WHERE pubname = 'migrate_pub'
+  ) THEN
+    DROP PUBLICATION migrate_pub;
+  END IF;
+END
+\$\$;
+SQL
+
+done
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/files/drop_subscriptions.sh

@@ -0,0 +1,29 @@
+#!/usr/bin/env bash
+set -e
+
+echo "== Dropping subscriptions =="
+
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "Processing DB: $db"
+
+  SUB="migrate_sub_${db}"
+
+  # Check if subscription exists
+  EXISTS=$(psql -d "$db" -Atqc "SELECT 1 FROM pg_subscription WHERE subname = '$SUB'")
+
+  if [ "$EXISTS" = "1" ]; then
+    echo "  Found subscription: $SUB"
+
+    # Disable first (good practice)
+    psql -d "$db" -c "ALTER SUBSCRIPTION $SUB DISABLE;"
+
+    # Drop it (must be top-level)
+    psql -d "$db" -c "DROP SUBSCRIPTION $SUB;"
+
+  else
+    echo "  No subscription: $SUB"
+  fi
+
+done
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/files/dump_all_databases.sh

@@ -0,0 +1,9 @@
+#!/bin/bash
+cd /tmp && \
+(pg_dumpall --globals-only > globals.sql) && \
+psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN (''template0'',''postgres'')" | \
+xargs -I{} -P4 sh -c "
+  pg_dump -Fd -j 4 -d \"{}\" -f dump_{} &&
+  tar -cf - dump_{} | zstd -19 -T0 > dump_{}.tar.zst &&
+  rm -rf dump_{}
+"

site-cookbooks/kosmos_postgresql/files/dump_database.sh

@@ -0,0 +1,10 @@
+#!/bin/bash
+set -euo pipefail
+
+DB_NAME="${1:?Usage: $0 <database_name>}"
+
+cd /tmp
+
+pg_dump -Fd -j 4 -d "$DB_NAME" -f "dump_${DB_NAME}"
+tar -cf - "dump_${DB_NAME}" | zstd -19 -T0 > "dump_${DB_NAME}.tar.zst"
+rm -rf "dump_${DB_NAME}"

site-cookbooks/kosmos_postgresql/files/fix_sequences.sh

@@ -0,0 +1,35 @@
+#!/bin/bash
+set -e
+
+DB="$1"
+
+if [ -z "$DB" ]; then
+  echo "Usage: $0 <database>"
+  exit 1
+fi
+
+echo "== Fixing sequences in database: $DB =="
+
+SQL=$(psql -d "$DB" -Atqc "
+  SELECT
+    'SELECT setval(' ||
+    quote_literal(pg_get_serial_sequence(quote_ident(n.nspname)||'.'||quote_ident(c.relname), a.attname)) ||
+    ', COALESCE(MAX(' || quote_ident(a.attname) || '), 0) + 1, false) FROM ' ||
+    quote_ident(n.nspname)||'.'||quote_ident(c.relname) || ';'
+  FROM pg_class c
+  JOIN pg_namespace n ON n.oid = c.relnamespace
+  JOIN pg_attribute a ON a.attrelid = c.oid
+  WHERE c.relkind = 'r'
+    AND a.attnum > 0
+    AND NOT a.attisdropped
+    AND pg_get_serial_sequence(quote_ident(n.nspname)||'.'||quote_ident(c.relname), a.attname) IS NOT NULL;
+")
+
+if [ -z "$SQL" ]; then
+  echo "No sequences found in $DB"
+  exit 0
+fi
+
+echo "$SQL" | psql -d "$DB"
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/files/fix_sequences_in_all_databases.sh

@@ -0,0 +1,38 @@
+#!/bin/bash
+set -e
+
+echo "== Fixing sequences across all databases =="
+
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "---- DB: $db ----"
+
+  # Generate fix statements
+  SQL=$(psql -d "$db" -Atqc "
+    SELECT
+      'SELECT setval(' ||
+      quote_literal(pg_get_serial_sequence(quote_ident(n.nspname)||'.'||quote_ident(c.relname), a.attname)) ||
+      ', COALESCE(MAX(' || quote_ident(a.attname) || '), 0) + 1, false) FROM ' ||
+      quote_ident(n.nspname)||'.'||quote_ident(c.relname) || ';'
+    FROM pg_class c
+    JOIN pg_namespace n ON n.oid = c.relnamespace
+    JOIN pg_attribute a ON a.attrelid = c.oid
+    WHERE c.relkind = 'r'
+      AND a.attnum > 0
+      AND NOT a.attisdropped
+      AND pg_get_serial_sequence(quote_ident(n.nspname)||'.'||quote_ident(c.relname), a.attname) IS NOT NULL;
+  ")
+
+  if [ -z "$SQL" ]; then
+    echo "No sequences found in $db"
+    continue
+  fi
+
+  echo "Fixing sequences in $db..."
+
+  # Execute generated statements
+  echo "$SQL" | psql -d "$db"
+
+done
+
+echo "== Done fixing sequences =="
+

site-cookbooks/kosmos_postgresql/files/list_publications.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "DB: $db"
+  psql -d "$db" -Atqc "SELECT pubname FROM pg_publication;"
+done

site-cookbooks/kosmos_postgresql/files/list_replication_slots.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+psql -c "
+SELECT slot_name,
+pg_size_pretty(pg_wal_lsn_diff(pg_current_wal_lsn(), restart_lsn))
+FROM pg_replication_slots;"

site-cookbooks/kosmos_postgresql/files/list_subscriptions.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "==== DB: $db ===="
+  psql -d "$db" -c "SELECT * FROM pg_stat_subscription;"
+done

site-cookbooks/kosmos_postgresql/files/restore_all_databases.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+set -euo pipefail
+
+cd /tmp
+
+for f in dump_*.tar.zst; do
+  db=$(echo $f | sed "s/dump_\(.*\)\.tar\.zst/\1/")
+  echo "Restoring $db"
+  zstd -d "$f" -c | tar -xf -
+  pg_restore -j 4 -d "$db" dump_$db
+  rm -rf "dump_$db"
+done

site-cookbooks/kosmos_postgresql/files/restore_database.sh

@@ -0,0 +1,14 @@
+#!/bin/bash
+set -euo pipefail
+
+DB_NAME="${1:?Usage: $0 <database_name>}"
+
+cd /tmp
+
+FILE="dump_${DB_NAME}.tar.zst"
+DIR="dump_${DB_NAME}"
+
+echo "Restoring $DB_NAME"
+zstd -d "$FILE" -c | tar -xf -
+pg_restore -j 4 -d "$DB_NAME" "$DIR"
+rm -rf "$DIR"

site-cookbooks/kosmos_postgresql/libraries/helpers.rb

@@ -36,10 +36,16 @@ class Chef
       end
     end
 
-    def postgresql_service_name
-      postgresql_version = "12"
+    def postgresql_version
+      node['kosmos_postgresql']['postgresql_version']
+    end
 
+    def postgresql_service
       "postgresql@#{postgresql_version}-main"
     end
+
+    def postgresql_data_dir
+      "/var/lib/postgresql/#{postgresql_version}/main"
+    end
   end
 end

site-cookbooks/kosmos_postgresql/recipes/management_scripts.rb

@@ -0,0 +1,121 @@
+#
+# Cookbook:: kosmos_postgresql
+# Recipe:: management_scripts
+#
+
+credentials = data_bag_item('credentials', 'postgresql')
+
+cookbook_file "/usr/local/bin/pg_dump_all_databases" do
+  source "dump_all_databases.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_dump_database" do
+  source "dump_database.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_restore_all_databases" do
+  source "restore_all_databases.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_restore_database" do
+  source "restore_database.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_create_replication_publications" do
+  source "create_publications.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_create_replication_publication" do
+  source "create_publication.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_drop_replication_publications" do
+  source "drop_publications.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_list_replication_publications" do
+  source "list_publications.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_list_replication_slots" do
+  source "list_replication_slots.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+template "/usr/local/bin/pg_create_replication_subscriptions" do
+  source "create_subscriptions.sh.erb"
+  user "postgres"
+  group "postgres"
+  mode "0740"
+  variables pg_host: "pg.kosmos.local",
+            pg_port: 5432,
+            pg_user: "replication",
+            pg_pass: credentials["replication_password"]
+  sensitive true
+end
+
+template "/usr/local/bin/pg_create_replication_subscription" do
+  source "create_subscription.sh.erb"
+  user "postgres"
+  group "postgres"
+  mode "0740"
+  variables pg_host: "pg.kosmos.local",
+            pg_port: 5432,
+            pg_user: "replication",
+            pg_pass: credentials["replication_password"]
+  sensitive true
+end
+
+cookbook_file "/usr/local/bin/pg_drop_replication_subscriptions" do
+  source "drop_subscriptions.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_list_replication_subscriptions" do
+  source "list_subscriptions.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_fix_sequences_in_all_databases" do
+  source "fix_sequences.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end
+
+cookbook_file "/usr/local/bin/pg_fix_sequences" do
+  source "fix_sequences.sh"
+  user "postgres"
+  group "postgres"
+  mode "0744"
+end

site-cookbooks/kosmos_postgresql/recipes/primary.rb

@@ -3,31 +3,6 @@
 # Recipe:: primary
 #
 
-postgresql_version = "12"
-postgresql_service = "postgresql@#{postgresql_version}-main"
-
-service postgresql_service do
-  supports restart: true, status: true, reload: true
-end
-
 postgresql_custom_server postgresql_version do
   role "primary"
 end
-
-postgresql_access "zerotier members" do
-  access_type "host"
-  access_db "all"
-  access_user "all"
-  access_addr "10.1.1.0/24"
-  access_method "md5"
-  notifies :reload, "service[#{postgresql_service}]", :immediately
-end
-
-postgresql_access "zerotier members replication" do
-  access_type "host"
-  access_db "replication"
-  access_user "replication"
-  access_addr "10.1.1.0/24"
-  access_method "md5"
-  notifies :reload, "service[#{postgresql_service}]", :immediately
-end

site-cookbooks/kosmos_postgresql/recipes/replica.rb

@@ -3,54 +3,31 @@
 # Recipe:: replica
 #
 
-postgresql_version = "12"
-postgresql_service = "postgresql@#{postgresql_version}-main"
-
 postgresql_custom_server postgresql_version do
   role "replica"
 end
 
-service postgresql_service do
-  supports restart: true, status: true, reload: true
-end
-
 postgresql_data_bag_item = data_bag_item('credentials', 'postgresql')
 
 primary = postgresql_primary
 
-unless primary.nil?
-  # TODO
-  postgresql_data_dir = "/var/lib/postgresql/#{postgresql_version}/main"
+if primary.nil?
+  Chef::Log.warn("No PostgreSQL primary node found. Skipping replication setup.")
+  return
+end
 
-  # FIXME get zerotier IP
-  execute "set up replication" do
-    command <<-EOF
+# TODO Replace pg.kosmos.local with private IP once available
+# via proper node attribute
+# https://gitea.kosmos.org/kosmos/chef/issues/263
+execute "set up replication" do
+  command <<-EOF
 systemctl stop #{postgresql_service}
 mv #{postgresql_data_dir} #{postgresql_data_dir}.old
 pg_basebackup -h pg.kosmos.local -U replication -D #{postgresql_data_dir} -R
 chown -R postgres:postgres #{postgresql_data_dir}
 systemctl start #{postgresql_service}
-    EOF
-    environment 'PGPASSWORD' => postgresql_data_bag_item['replication_password']
-    sensitive true
-    not_if { ::File.exist? "#{postgresql_data_dir}/standby.signal" }
-  end
-
-  postgresql_access "zerotier members" do
-    access_type "host"
-    access_db "all"
-    access_user "all"
-    access_addr "10.1.1.0/24"
-    access_method "md5"
-    notifies :reload, "service[#{postgresql_service}]", :immediately
-  end
-
-  postgresql_access "zerotier members replication" do
-    access_type "host"
-    access_db "replication"
-    access_user "replication"
-    access_addr "10.1.1.0/24"
-    access_method "md5"
-    notifies :reload, "service[#{postgresql_service}]", :immediately
-  end
+  EOF
+  environment 'PGPASSWORD' => postgresql_data_bag_item['replication_password']
+  sensitive true
+  not_if { ::File.exist? "#{postgresql_data_dir}/standby.signal" }
 end

site-cookbooks/kosmos_postgresql/recipes/replica_logical.rb

@@ -0,0 +1,15 @@
+#
+# Cookbook:: kosmos_postgresql
+# Recipe:: replica_logical
+#
+
+postgresql_custom_server postgresql_version do
+  role "replica_logical"
+end
+
+# primary = postgresql_primary
+#
+# if primary.nil?
+#   Chef::Log.warn("No PostgreSQL primary node found. Skipping replication setup.")
+#   return
+# end

site-cookbooks/kosmos_postgresql/resources/server.rb

@@ -56,13 +56,15 @@ action :create do
     timezone: "UTC", # default is GMT
     listen_addresses: "0.0.0.0",
     promote_trigger_file: "#{postgresql_data_dir}/failover.trigger",
-    wal_keep_segments: 256
+    wal_level: "logical",
+    wal_keep_size: 4096, # 256 segments, 16MB each
+    max_replication_slots: 16
   }
 
   postgresql_server_conf "main" do
     version postgresql_version
     additional_config additional_config
-    notifies :reload, "service[#{postgresql_service}]", :delayed
+    notifies :restart, "service[#{postgresql_service}]", :delayed
   end
 
   postgresql_user "replication" do
@@ -70,6 +72,24 @@ action :create do
     replication true
     password postgresql_credentials['replication_password']
   end
+
+  postgresql_access "all members" do
+    access_type "host"
+    access_db "all"
+    access_user "all"
+    access_addr node['kosmos_postgresql']['access_addr']
+    access_method "md5"
+    notifies :reload, "service[#{postgresql_service}]", :immediately
+  end
+
+  postgresql_access "replication members" do
+    access_type "host"
+    access_db "replication"
+    access_user "replication"
+    access_addr node['kosmos_postgresql']['access_addr']
+    access_method "md5"
+    notifies :reload, "service[#{postgresql_service}]", :immediately
+  end
 end
 
 action_class do

site-cookbooks/kosmos_postgresql/templates/create_subscription.sh.erb

@@ -0,0 +1,31 @@
+#!/bin/bash
+set -euo pipefail
+
+DB_NAME="${1:?Usage: $0 <database_name>}"
+
+echo "== Processing DB: $DB_NAME =="
+
+SLOT="migrate_slot_${DB_NAME}"
+SUB="migrate_sub_${DB_NAME}"
+
+psql -d "$DB_NAME" -v ON_ERROR_STOP=1 <<SQL
+DO \$\$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_subscription WHERE subname = '$SUB'
+  ) THEN
+    CREATE SUBSCRIPTION $SUB
+    CONNECTION 'host=<%= @pg_host %> port=<%= @pg_port %> dbname=$DB_NAME user=<%= @pg_user %> password=<%= @pg_pass %>'
+    PUBLICATION migrate_pub
+    WITH (
+      slot_name = '$SLOT',
+      create_slot = false,
+      copy_data = false,
+      enabled = true
+    );
+  END IF;
+END
+\$\$;
+SQL
+
+echo "== Done =="

site-cookbooks/kosmos_postgresql/templates/create_subscriptions.sh.erb

@@ -0,0 +1,34 @@
+#!/bin/bash
+set -e
+
+echo "== Creating subscriptions for all databases =="
+
+for db in $(psql -Atqc "SELECT datname FROM pg_database WHERE datallowconn AND datname NOT IN ('template1','postgres')"); do
+  echo "Processing DB: $db"
+
+  SLOT="migrate_slot_${db}"
+  SUB="migrate_sub_${db}"
+
+  psql -d "$db" -v ON_ERROR_STOP=1 <<SQL
+DO \$\$
+BEGIN
+  IF NOT EXISTS (
+    SELECT 1 FROM pg_subscription WHERE subname = '$SUB'
+  ) THEN
+    CREATE SUBSCRIPTION $SUB
+    CONNECTION 'host=<%= @pg_host %> port=<%= @pg_port %> dbname=$db user=<%= @pg_user %> password=<%= @pg_pass %>'
+    PUBLICATION migrate_pub
+    WITH (
+      slot_name = '$SLOT',
+      create_slot = false,
+      copy_data = false,
+      enabled = true
+    );
+  END IF;
+END
+\$\$;
+SQL
+
+done
+
+echo "== Done =="