Use data dir and remove /tmp/pg_* caches (#100)

Go back to using `/var/lib/postgresql/data` instead of
`/var/lib/postgresql/tembo` because, although it made it compatible with
stock CNPG, it was incompatible with Tembo Cloud, which also relies on
CNPG, so it would never be mounted. It means the image will no longe be
compatible with stock CNPG, sadly, because
`/var/lib/postgresql/data/share` is still the shared directory, and CNPG
will wipe it out when it mounts the `pgdata` volume at
`/var/lib/postgresql/data`.

Also remove the copying of the contents of `/var/lib/postgresql/data` to
`/tmp/pg_sharedir` and `/tmp/pg_pkglibdir` from all images. As of
tembo-io/tembo#1185, the init pod now copies files directly from
`/var/lib/postgresql/data`.

Author: theory

Diff for: CONTAINER_README.md

@@ -6,29 +6,20 @@ live here in `/var/lib/postgresql`. Here's a description of subdirectories and
 their purposes:
 
 *   `data`: The directory in which the Docker entrypoint initializes
-    a cluster and which [CloudNativePG] mounts as a persistent volume.
+    a cluster and which [Tembo Cloud] mounts as a persistent volume.
 *   `data/pgdata`: The data directory initialized by the Docker entrypoint and
-    [CloudNativePG].
-*   `tembo`: The directory that [Tembo Cloud] mounts as the persistent volume
-    for the database and extensions.
-*   `tembo/pgdata`: The data directory initialized by [Tembo Cloud].
-*   `tembo/lib`: A directory for shared library files required by extensions.
-*   `tembo/mod`: A directory for extension module library files.
-*   `tembo/share`: The directory for architecture-independent support files
+    [Tembo Cloud].
+*   `data/lib`: A directory for shared library files required by extensions.
+*   `data/mod`: A directory for extension module library files.
+*   `data/share`: The directory for architecture-independent support files
     used by Postgres. This is the directory used by `pg_config --sharedir` to
-    install non-binary extension files. Its files are copied from
-    `/tmp/pg_sharedir` when the Tembo operator initializes the volume.
-
+    install non-binary extension files.
 
 Other useful locations around the system:
 
 *   `/usr/lib/postgresql`: The base directory for Postgres itself.
 *   `/usr/local/bin/docker-entrypoint.sh`: The docker entrypoint script, which
     initializes and starts a Postgres cluster in `/var/lib/postgresql/data/pgdata`.
-*   `/etc/ld.so.conf.d/postgres.conf`: Configures the library loader to look
-    for library files in the Postgres library directory (`pg_config --libdir`).
-*   `/etc/ld.so.conf.d/tembo.conf`: Configures the library loader to look
-    for library files in `/var/lib/postgresql/tembo/${OS_NAME}/lib`.
 
   [CloudNativePG]: https://cloudnative-pg.io
   [Tembo Cloud]: https://tembo.io/docs/product/cloud/overview "Tembo Cloud Overview"

Diff for: Dockerfile

@@ -5,20 +5,19 @@ ARG PG_VERSION
 ARG PG_MAJOR=${PG_VERSION%%.*}
 ARG PG_PREFIX=/usr/lib/postgresql
 ARG PG_HOME=/var/lib/postgresql
-ARG CNPG_VOLUME=${PG_HOME}/data
+ARG DATA_VOLUME=${PG_HOME}/data
 
 # Tembo-specific volume mount, sharedir, dynamic_library_path target, and
 # System LLD dir.
-ARG TEMBO_VOLUME=${PG_HOME}/tembo
-ARG TEMBO_SHARE_DIR=${TEMBO_VOLUME}/share
-ARG TEMBO_PG_LIB_DIR=${TEMBO_VOLUME}/mod
-ARG TEMBO_LD_LIB_DIR=${TEMBO_VOLUME}/lib
+ARG TEMBO_SHARE_DIR=${DATA_VOLUME}/share
+ARG TEMBO_PG_MOD_DIR=${DATA_VOLUME}/mod
+ARG TEMBO_LD_LIB_DIR=${DATA_VOLUME}/lib
 
 # Set rpath to search the Postgres lib directory, then the Tembo Postgres lib
 # directory, where Trunk-installed extension libraries will live, and the
 # Tembo OS lib directory, where Trunk-installed third-party libraries will
 # live. https://lekensteyn.nl/rpath.html
-ARG TEMBO_RPATH=${PG_PREFIX}/lib:${TEMBO_PG_LIB_DIR}:${TEMBO_LD_LIB_DIR}
+ARG TEMBO_RPATH=${PG_PREFIX}/lib:${TEMBO_PG_MOD_DIR}:${TEMBO_LD_LIB_DIR}
 
 ##############################################################################
 # Build trunk.
@@ -107,7 +106,7 @@ RUN set -ex; \
 ##############################################################################
 # Install additional stuff for the dev image.
 FROM build AS dev-install
-ARG PG_PREFIX PG_HOME TEMBO_LD_LIB_DIR TEMBO_PG_LIB_DIR CNPG_VOLUME
+ARG PG_PREFIX PG_HOME TEMBO_LD_LIB_DIR TEMBO_PG_MOD_DIR
 
 ENV DEBIAN_FRONTEND=noninteractive
 RUN set -ex; \
@@ -121,7 +120,7 @@ RUN set -ex; \
     apt-get clean -y; \
     localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8; \
     rm -rf /var/lib/apt/lists/* /var/cache/* /var/log/*; \
-    mkdir -p "${TEMBO_LD_LIB_DIR}" "${TEMBO_PG_LIB_DIR}" "${CNPG_VOLUME}"; \
+    mkdir -p "${TEMBO_LD_LIB_DIR}" "${TEMBO_PG_MOD_DIR}"; \
     groupadd -r postgres --gid=999 && \
 	useradd -r -g postgres --uid=26 --home-dir=${PG_HOME} --shell=/bin/bash postgres && \
     chown -R postgres:postgres ${PG_HOME};
@@ -144,7 +143,7 @@ ENTRYPOINT ["docker-entrypoint.sh"]
 ##############################################################################
 # Install the dependencies necessary for the base image.
 FROM ${BASE} AS install
-ARG CNPG_VOLUME PACKAGES TEMBO_LD_LIB_DIR TEMBO_PG_LIB_DIR PG_PREFIX PG_HOME
+ARG PACKAGES TEMBO_LD_LIB_DIR TEMBO_PG_MOD_DIR PG_PREFIX PG_HOME
 
 # Copy the PostgreSQL files and trunk.
 COPY --link --from=build --parents /var/lib/./postgresql /var/lib/
@@ -173,14 +172,9 @@ RUN apt-get update && apt-get upgrade -y && apt-get install --no-install-recomme
 ENV PATH=${PG_PREFIX}/bin:$PATH
 RUN set -xe; \
     apt-get clean -y; \
-    # Set up en_US.UTF-8
     localedef -i en_US -c -f UTF-8 -A /usr/share/locale/locale.alias en_US.UTF-8; \
-    mkdir -p "${TEMBO_LD_LIB_DIR}" "${TEMBO_PG_LIB_DIR}" "${CNPG_VOLUME}"; \
-    rm -rf /var/lib/apt/lists/* /var/cache/* /var/log/*; \
-    # Stash away sharedir and etc so the Tembo operator can copy them back
-    # when it initializes the pod. (We should be able to do it during pod
-    # initialization without a temp copy: https://stackoverflow.com/a/72269316/79202)
-    cp -lr "$(pg_config --sharedir)" /tmp/pg_sharedir;
+    mkdir -p "${TEMBO_LD_LIB_DIR}" "${TEMBO_PG_MOD_DIR}"; \
+    rm -rf /var/lib/apt/lists/* /var/cache/* /var/log/*;
 
 # Add the README and entrypoint script.
 COPY CONTAINER_README.md "${PG_HOME}/README.md"

Diff for: README.md

@@ -14,19 +14,6 @@ Ubuntu Noble (24.04) and Jimmy (22.04) for the ARM64 and AMD64 processors.
     docker exec -it tembo-postgres psql
     ```
 
-*   Runs in [CloudNativePG]; just set the `imageName` key in the `spec`
-    section of the `Cluster` manifest:
-
-    ```yaml
-    apiVersion: postgresql.cnpg.io/v1
-    kind: Cluster
-    metadata:
-      # [...]
-    spec:
-      imageName: quay.io/tembo/postgres:17
-      #[...]
-    ```
-
 *   Based on the latest and previous LTS Ubuntu [Ubuntu Linux], currently
     24.04 LTS "Noble Numbat" and 22.04 LTS "Jammy Jellyfish".
 
@@ -41,10 +28,13 @@ Ubuntu Noble (24.04) and Jimmy (22.04) for the ARM64 and AMD64 processors.
 ## Building
 
 The easiest way to build and load the `postgres` and `postgres-dev` images
-into Docker is:
+into Docker is to set up a temporary registry, push to it, then pull from it:
 
 ```sh
-arch="$(uname -m)" pg_version=17.4 docker buildx bake --load
+docker run -d -p 5001:5000 --restart=always --name registry registry:2
+registry=localhost:5001 arch="$(uname -m)" pg_version=17.4 docker buildx bake --push
+docker pull localhost:5001/postgres
+docker pull localhost:5001/postgres-dev
 ```
 
 Set these environment variables to modify the build behavior:
@@ -152,12 +142,9 @@ And an image built on the latest Postgres includes the tag:
 *   `/var/lib/postgresql`: The home directory for the `postgres` user where
     all the potentially persistent data files and libraries live.
 
-*   `/var/lib/postgresql/data`: The default data directory where the Docker
-    entrypoint script and [CloudNativePG] store the data files in a `pgdata`
-    subdirectory. Mount a volume to this directory for data persistence.
-
-*   `/var/lib/postgresql/tembo`: The directory where [Tembo Cloud] mounts a
-    persistent volume and stores persistent data:
+*   `/var/lib/postgresql/data`: The directory where the Docker entrypoint
+    script initializes the database in the `pgdata` subdirectory, and where
+    [Tembo Cloud] mounts a persistent volume and stores persistent data:
     *   `pgdata`: Tembo initializes and runs the cluster from this
         subdirectory.
     *   `mod`: Tembo stores extension module libraries this subdirectory.
@@ -166,7 +153,13 @@ And an image built on the latest Postgres includes the tag:
         subdirectory.
 
 *   `/usr/lib/postgresql`: The home of the PostgreSQL binaries, libraries, and
-    header & locale files. Immutable in [CloudNativePG] and [Tembo Cloud].
+    header & locale files. Immutable in [Tembo Cloud].
+
+### Cloud Native Postgres Support
+
+Unfortunately, this image does not work on [CloudNativePG]. This is because it
+currently stores files in `/var/lib/postgresql/data` that [CloudNativePG]
+masks when it mounts the directory as a volume.
 
 ## Tasks
 

Diff for: geo-cnpg/Dockerfile

@@ -45,11 +45,7 @@ RUN /usr/bin/trunk install postgis --version 3.5.0 \
     && make -C MobilityDB/build -j8 install \
     && rm -rf MobilityDB
 
-# cache all extensions
-RUN set -eux; \
-    ldconfig; \
-    cp -r $(pg_config --pkglibdir)/* /tmp/pg_pkglibdir; \
-    cp -r $(pg_config --sharedir)/* /tmp/pg_sharedir;
+RUN ldconfig
 
 ENV LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH
 ENV XDG_CACHE_HOME=/var/lib/postgresql/data/tembo/.cache

Diff for: ml-cnpg/Dockerfile

@@ -19,11 +19,6 @@ RUN set -eux; \
   trunk install jsonb_plpython3u; \
   trunk install ltree_plpython3u;
 
-# cache all extensions
-RUN set -eux; \
-      cp -r $(pg_config --pkglibdir)/* /tmp/pg_pkglibdir; \
-      cp -r $(pg_config --sharedir)/* /tmp/pg_sharedir;
-
 # Revert the postgres user to id 26
 RUN usermod -u 26 postgres
 USER 26

Diff for: standard-cnpg/Dockerfile

@@ -122,11 +122,7 @@ RUN set -eux; \
     ldconfig; \
     # Install auto_explain and pg_stat_statements.
     /usr/bin/trunk install auto_explain; \
-    /usr/bin/trunk install pg_stat_statements; \
-    # cache pg_stat_statements and auto_explain and pg_stat_kcache to temp directory
-    mkdir /tmp/pg_pkglibdir /tmp/pg_sharedir; \
-    cp -r $(pg_config --pkglibdir)/* /tmp/pg_pkglibdir; \
-    cp -r $(pg_config --sharedir)/* /tmp/pg_sharedir
+    /usr/bin/trunk install pg_stat_statements;
 
 # Revert the postgres user to id 26
 RUN usermod -u 26 postgres

Diff for: tembo-pg-cnpg/Dockerfile

@@ -39,13 +39,6 @@ RUN trunk install pg_stat_statements
 # Install auto_explain
 RUN trunk install auto_explain
 
-# cache pg_stat_statements and auto_explain and pg_stat_kcache to temp directory
-RUN set -eux; \
-	  mkdir /tmp/pg_pkglibdir; \
-	  mkdir /tmp/pg_sharedir; \
-      cp -r $(pg_config --pkglibdir)/* /tmp/pg_pkglibdir; \
-      cp -r $(pg_config --sharedir)/* /tmp/pg_sharedir
-
 # Revert the postgres user to id 26
 RUN usermod -u 26 postgres
 USER 26