Outline Nominatim Deployment

Nominatim.md

@@ -0,0 +1,25 @@
+# Deploying Nominatim
+The [unofficial Helm Charts](https://nominatim.org/release-docs/latest/admin/Installation/) come with 
+(documentation here)[https://github.com/robjuz/helm-charts/blob/master/charts/nominatim/README.md].
+The instructions here are based upon that documentation.
+> The cluster will need at least 4GB of available memory (on the node where this
+> is deployed) to perform initialization. If this isn't possible, update 
+> `init_values.yaml` to use less `maintenance_work_mem` and consider updating 
+> `pbfUrl` to use a smaller region instead of the global data.
+
+## Initializing the Nominatim Services
+Some default values files are included with Diana.
+To initialize Nominatim with global data to its own namespace:
+```shell
+helm upgrade -n nominatim nominatim robjuz/nominatim -f neon_diana_utils/nominatim/init_values.yaml
+```
+
+This will likely take on the order of hours to complete. You may use the Dashboard
+or k9s to monitor the job `nominatim-init` (in the `nominatim` namespace). If the
+job is interrupted, it is possible that the database is left in a corrupted state;
+it is recommended to `kubectl delete namespace nominatim` and start over if the
+init process is interrupted.
+> Note: The Kubernetes Dashboard may show the `nominatim-init` job as having 1
+> completion before the job has completed. `k9s` does not appear to have this bug.
+> If you can't use `k9s` then consider watching the Job logs to make sure the init
+> actually completes.

neon_diana_utils/nominatim/deploy_values.yaml

@@ -0,0 +1,7 @@
+  nominatimInitialize:
+    enabled: false
+
+  postgresql:
+    primary:
+      persistence:
+        size: 200Gi

neon_diana_utils/nominatim/init_values.yaml

@@ -0,0 +1,24 @@
+  nominatimInitialize:
+    enabled: true
+    pbfUrl: "https://ftpmirror.your.org/pub/openstreetmap/pbf/planet-latest.osm.pbf"
+    importWikipedia: True
+    resources:
+      requests:
+        memory: "4Gi"
+
+  postgresql:
+    primary:
+      persistence:
+        size: 200Gi
+      extendedConfiguration: |
+        shared_buffers = 2GB
+        maintenance_work_mem = 2GB
+        autovacuum_work_mem = 2GB
+        work_mem = 50MB
+        effective_cache_size = 24GB
+        synchronous_commit = off
+        max_wal_size = 2GB
+        checkpoint_timeout = 10min
+        checkpoint_completion_target = 0.9
+        fsync = off
+        full_page_writes = off