Postgres README improvements

Signed-off-by: Irving Popovetsky <[email protected]>

postgresql/README.md

@@ -12,7 +12,7 @@ hab start core/postgresql
 ```
 or
 ```
-docker run starkandwayne/postgresql
+docker run chefdemo/postgresql
 ```
 if you want to bring up the pre-exported docker image.
 
@@ -22,7 +22,7 @@ This plan supports running clustered PostgreSQL by utilizing Habitat's native le
 
 You can run an example cluster via docker-compose after exporting a docker container from this plan:
 ```
-$ hab pkg export docker core/postgresql
+$ hab pkg export docker $(ls -1t results/*.hart | head -1)
 ```
 
 The docker post-process should create a docker image named `core/postgresql` and it should be available on your local machine
@@ -33,22 +33,35 @@ version: '3'
 
 services:
   pg1:
-    image: core/postgresql
-    command: "start core/postgresql --group cluster --topology leader"
+    image: chefdemo/postgresql
+    command: --group cluster
+      --topology leader
+    volumes:
+      - pg1-data:/hab/svc/postgresql/data
   pg2:
-    image: core/postgresql
-    command: "start core/postgresql --group cluster --topology leader --peer pg1"
+    image: chefdemo/postgresql
+    command: --group cluster
+      --topology leader
+      --peer pg1
+    volumes:
+      - pg2-data:/hab/svc/postgresql/data
   pg3:
-    image: core/postgresql
-    command: "start core/postgresql --group cluster --topology leader --peer pg1"
+    image: chefdemo/postgresql
+    command: --group cluster
+      --topology leader
+      --peer pg1
+    volumes:
+      - pg3-data:/hab/svc/postgresql/data
+
+volumes:
+  pg1-data:
+  pg2-data:
+  pg3-data:
 EOF
 
 docker-compose up
 ```
 
-We intend to support a prod ready clustering solution similar to [patroni](https://github.com/zalando/patroni) and are working on getting there quickly.
-
-
 ## Binding
 
 Consuming services can bind to PostgreSQL via:
@@ -59,13 +72,18 @@ hab start <origin>/<app> --bind database:postgresql.default --peer <pg-host>
 
 Superuser access is exposed to consumers when binding and we advise that required databases, schemas and roles be created and migrations get run in the init hook of the consuming service. The created roles (with restricted access) should then be exposed to the application code that will get run.
 
-To support binding to either standalone or custered PostgreSQL services we suggest using the `.first` field of the binding in the handlebars interpolation:
+To support binding to either standalone or custered PostgreSQL services we suggest using the `eachAlive` and `@first` helpers of the binding in the handlebars interpolation:
 ```
-{{#with bind.database.first as |pg| }}
-PGPASSWORD={{pg.cfg.superuser_password}}
-PGUSER={{pg.cfg.superuser_name}}
-PGHOST={{pg.sys.ip}}
-{{/with}}
+{{~ #if bind.database}}
+  {{~ #eachAlive bind.database.members as |member|}}
+    {{~ #if @first}}
+PGHOST="{{member.sys.ip}}"
+PGPORT="{{member.cfg.port}}"
+PGUSER="{{member.cfg.superuser_name}}"
+PGPASSWORD="{{member.cfg.superuser_password}}"
+    {{~ /if}}
+  {{~ /eachAlive}}
+{{~ /if}}
 ```
 
-`.first` will always point to the leader when present.
+In clustered situations, there is no reliable way to determine the leader via template helpers.  However, in census data the members are ordered chronologically by join time.