Prepopulating PostGIS with OSM data works with Docker Desktop, but not with Azure pipeline

Question

When I build the Dockerfile below by running the command manually at a Windows notebook

docker build -f Dockerfile --no-cache --tag my-postgis:16-3.4 \
    --build-arg DOWNLOAD_URLS="https://download.geofabrik.de/europe/germany/berlin-latest.osm.pbf https://download.geofabrik.de/europe/germany/hamburg-latest.osm.pbf" \
    --build-arg OSM_PASSWORD=osm_password \
    --build-arg POSTGIS_TAG=16-3.4 .

then it works as expected and the prefilled PostgreSQL database is there in the container launched in Docker Desktop:

But when I build the same Dockerfile by an Azure pipeline:

- task: AzureCLI@2
  displayName: Build and push PostGIS image
  inputs:
    azureSubscription: $(ArmConnection)
    scriptType: bash
    scriptLocation: inlineScript
    inlineScript: |
      image_tag='$(ContainerRegistry)/postgis-${{ parameters.OsmRegion }}:${{ parameters.PostGisTag }}'
      # delete old images to avoid the error "no space left on device"
      docker system prune --all --force
      docker build --file $(Build.SourcesDirectory)/src/Services/PostGIS/Dockerfile --no-cache --tag $image_tag --build-arg POSTGIS_TAG=${{parameters.PostGisTag}} --build-arg OSM_PASSWORD=$(OsmPassword) --build-arg DOWNLOAD_URLS="$(DownloadUrls)" $(Build.SourcesDirectory)/src/
      # building the Docker image takes longer time, so be sure to acr login just before pushing the image
      az acr login -n $(ContainerRegistry)
      docker push $image_tag

and then pull the built image from ACR and launch it at Kubernetes, then the database folder is empty:

I understand that it is probably not a recommended practice to launch and stop a service during the "docker build".

The reason I am not populating the database by running a script in /docker-entrypoint-initdb.d folder is that in production we use the 28 GB big europe-latest.osm.pbf (and not the small Berlin file) and thus launching a pod would take several hours.

Below is my Dockerfile, I have created it by looking at the https://github.com/postgis/docker-postgis/blob/master/Dockerfile.alpine.template

ARG POSTGIS_TAG
FROM postgis/postgis:$POSTGIS_TAG
EXPOSE 5432

# Install ps, top, netstat, curl, wget, osmium, osm2pgsql
RUN apt-get update && \
    apt-get upgrade -y && \
    apt install -y procps net-tools curl wget osmium-tool osm2pgsql

WORKDIR /data
RUN chown -R postgres:postgres /data
USER postgres

# Download one or multiple OSM files
ARG DOWNLOAD_URLS
RUN wget $DOWNLOAD_URLS
# Print summaries of the downloaded OSM files
RUN for f in *.osm.pbf; do osmium fileinfo -e "$f"; done
# Merge one or multiple files into a new file map.osm.pbf
RUN osmium merge *.osm.pbf -o map.osm.pbf
# Remove all files except the map.osm.pbf
RUN find . -type f -name '*.osm.pbf' | grep -v 'map.osm.pbf' | xargs rm -f

ARG OSM_PASSWORD
ENV PGPASSWORD=$OSM_PASSWORD
ENV PGUSER=osm_user
ENV PGDATABASE=osm_database
ENV PGDATA=/var/lib/postgresql/data

# create PostgreSQL instance in the $PGDATA folder
# configure PostgreSQL as recommended by osm2pgsql doc
# start PostgreSQL and create database and user
# load data from the map.osm.pbf file into the database
# remove the map.osm.pbf file and stop PostgreSQL
# configure password based access for osm_user
RUN set -eux && \
    pg_ctl init && \
    echo "shared_buffers = 1GB"                >> $PGDATA/postgresql.conf && \
    echo "work_mem = 50MB"                     >> $PGDATA/postgresql.conf && \
    echo "maintenance_work_mem = 10GB"         >> $PGDATA/postgresql.conf && \
    echo "autovacuum_work_mem = 2GB"           >> $PGDATA/postgresql.conf && \
    echo "wal_level = minimal"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_timeout = 60min"          >> $PGDATA/postgresql.conf && \
    echo "max_wal_size = 10GB"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_completion_target = 0.9"  >> $PGDATA/postgresql.conf && \
    echo "max_wal_senders = 0"                 >> $PGDATA/postgresql.conf && \
    echo "random_page_cost = 1.0"              >> $PGDATA/postgresql.conf && \
    echo "password_encryption = scram-sha-256" >> $PGDATA/postgresql.conf && \
    pg_ctl start && \ 
    createuser --username=postgres $PGUSER && \
    createdb --username=postgres --encoding=UTF8 --owner=$PGUSER $PGDATABASE && \
    psql --username=postgres $PGDATABASE --command="ALTER USER $PGUSER WITH PASSWORD '$PGPASSWORD';" && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS postgis;' && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS hstore;' && \
    osm2pgsql --username=$PGUSER --database=$PGDATABASE --create --cache=60000 --hstore --latlong /data/map.osm.pbf && \
    rm -f /data/map.osm.pbf && \
    pg_ctl stop && \
    echo '# TYPE DATABASE USER ADDRESS METHOD'                > $PGDATA/pg_hba.conf && \
    echo "local all postgres peer"                           >> $PGDATA/pg_hba.conf && \
    echo "local $PGDATABASE $PGUSER           scram-sha-256" >> $PGDATA/pg_hba.conf && \
    echo "host  $PGDATABASE $PGUSER 0.0.0.0/0 scram-sha-256" >> $PGDATA/pg_hba.conf

I have also asked the same question at docker-postgis Github discussions.

Answer 1

I have found a solution for the empty /var/lib/postgresql/data folder, when building a postgis/postgis:16.3 based Docker image with prefilled OSM database and would like to share it here -

After changing just a single line my Dockerfile has built successfully:

ENV PGDATA=/postgis

The reason seems to be that the parent official postgres Dockerfile contains this conflicting line:

VOLUME /var/lib/postgresql/data

and thus the folder I was trying to use originally was cleared.

Below is my working Dockerfile (and yes I understand that the built image is too big, especially with Europe data and thus I have to improve it further and possibly use VOLUME or some Azure storage there):

ARG POSTGIS_TAG
FROM postgis/postgis:$POSTGIS_TAG
EXPOSE 5432

# Install ps, top, netstat, curl, wget, osmium, osm2pgsql
RUN apt-get update && \
    apt-get upgrade -y && \
    apt install -y procps net-tools curl wget osmium-tool osm2pgsql

# place PostGIS database into /postgis folder
WORKDIR /postgis
RUN chown -R postgres:postgres /postgis
# download OSM data files into /data folder
WORKDIR /data
RUN chown -R postgres:postgres /data
# drop root privileges
USER postgres

# Download one or multiple OSM files
ARG DOWNLOAD_URLS
RUN wget $DOWNLOAD_URLS
# Print summaries of the downloaded OSM files
RUN for f in *.osm.pbf; do osmium fileinfo -e "$f"; done
# Merge one or multiple files into a new file map.osm.pbf
RUN osmium merge *.osm.pbf -o map.osm.pbf
# Remove all files except the map.osm.pbf
RUN find . -type f -name '*.osm.pbf' | grep -v 'map.osm.pbf' | xargs rm -f

ARG OSM_PASSWORD
ENV PGPASSWORD=$OSM_PASSWORD
ENV PGUSER=osm_user
ENV PGDATABASE=osm_database
# do not use the /var/lib/postgresql/data folder here!
ENV PGDATA=/postgis

# create PostgreSQL instance in the $PGDATA folder
# configure PostgreSQL as recommended by osm2pgsql doc
# start PostgreSQL and create database and user
# load data from the map.osm.pbf file into the database
# remove the map.osm.pbf file and stop PostgreSQL
# configure password based access for osm_user
RUN set -eux && \
    pg_ctl init && \
    echo "shared_buffers = 1GB"                >> $PGDATA/postgresql.conf && \
    echo "work_mem = 50MB"                     >> $PGDATA/postgresql.conf && \
    echo "maintenance_work_mem = 10GB"         >> $PGDATA/postgresql.conf && \
    echo "autovacuum_work_mem = 2GB"           >> $PGDATA/postgresql.conf && \
    echo "wal_level = minimal"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_timeout = 60min"          >> $PGDATA/postgresql.conf && \
    echo "max_wal_size = 10GB"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_completion_target = 0.9"  >> $PGDATA/postgresql.conf && \
    echo "max_wal_senders = 0"                 >> $PGDATA/postgresql.conf && \
    echo "random_page_cost = 1.0"              >> $PGDATA/postgresql.conf && \
    echo "password_encryption = scram-sha-256" >> $PGDATA/postgresql.conf && \
    pg_ctl start && \ 
    createuser --username=postgres $PGUSER && \
    createdb --username=postgres --encoding=UTF8 --owner=$PGUSER $PGDATABASE && \
    psql --username=postgres $PGDATABASE --command="ALTER USER $PGUSER WITH PASSWORD '$PGPASSWORD';" && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS postgis;' && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS hstore;' && \
    osm2pgsql --username=$PGUSER --database=$PGDATABASE --create --cache=60000 --hstore --latlong /data/map.osm.pbf && \
    rm -f /data/map.osm.pbf && \
    pg_ctl stop && \
    echo '# TYPE DATABASE USER ADDRESS METHOD'                > $PGDATA/pg_hba.conf && \
    echo "local all postgres peer"                           >> $PGDATA/pg_hba.conf && \
    echo "local $PGDATABASE $PGUSER           scram-sha-256" >> $PGDATA/pg_hba.conf && \
    echo "host  $PGDATABASE $PGUSER 0.0.0.0/0 scram-sha-256" >> $PGDATA/pg_hba.conf

Example build command:

docker build -f Dockerfile --no-cache --tag my-postgis:16-3.4 \
    --build-arg DOWNLOAD_URLS="https://download.geofabrik.de/europe/germany/berlin-latest.osm.pbf https://download.geofabrik.de/europe/germany/hamburg-latest.osm.pbf" \
    --build-arg OSM_PASSWORD=osm_password \
    --build-arg POSTGIS_TAG=16-3.4 .

Answer 2

Checked with same Dockerfile with Ubuntu-latest agent on my side, it will reports Out of memory error:

After i reduce 60000 to 10000 for the command in Dockerfile, the pipeline is working fine on my side, the output of RUN ls -al $PGDATA is correct.

The Dockerfile:

ARG POSTGIS_TAG
FROM postgis/postgis:$POSTGIS_TAG
EXPOSE 5432

# Install ps, top, netstat, curl, wget, osmium, osm2pgsql
RUN apt-get update && \
    apt-get upgrade -y && \
    apt install -y procps net-tools curl wget osmium-tool osm2pgsql

WORKDIR /data
RUN chown -R postgres:postgres /data
USER postgres

# Download one or multiple OSM files
ARG DOWNLOAD_URLS

RUN wget --no-verbose $DOWNLOAD_URLS
# Print summaries of the downloaded OSM files

RUN for f in *.osm.pbf; do osmium fileinfo -e "$f"; done

# Merge one or multiple files into a new file map.osm.pbf
RUN osmium merge *.osm.pbf -o map.osm.pbf
# Remove all files except the map.osm.pbf

RUN find . -type f -name '*.osm.pbf' | grep -v 'map.osm.pbf' | xargs rm -f

ARG OSM_PASSWORD
ENV PGPASSWORD=$OSM_PASSWORD
ENV PGUSER=osm_user
ENV PGDATABASE=osm_database
ENV PGDATA=/var/lib/postgresql/data

RUN set -eux && \
    pg_ctl init && \
    echo "shared_buffers = 1GB"                >> $PGDATA/postgresql.conf && \
    echo "work_mem = 50MB"                     >> $PGDATA/postgresql.conf && \
    echo "maintenance_work_mem = 10GB"         >> $PGDATA/postgresql.conf && \
    echo "autovacuum_work_mem = 2GB"           >> $PGDATA/postgresql.conf && \
    echo "wal_level = minimal"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_timeout = 60min"          >> $PGDATA/postgresql.conf && \
    echo "max_wal_size = 10GB"                 >> $PGDATA/postgresql.conf && \
    echo "checkpoint_completion_target = 0.9"  >> $PGDATA/postgresql.conf && \
    echo "max_wal_senders = 0"                 >> $PGDATA/postgresql.conf && \
    echo "random_page_cost = 1.0"              >> $PGDATA/postgresql.conf && \
    echo "password_encryption = scram-sha-256" >> $PGDATA/postgresql.conf && \
    pg_ctl start && \ 
    createuser --username=postgres $PGUSER && \
    createdb --username=postgres --encoding=UTF8 --owner=$PGUSER $PGDATABASE && \
    psql --username=postgres $PGDATABASE --command="ALTER USER $PGUSER WITH PASSWORD '$PGPASSWORD';" && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS postgis;' && \
    psql --username=postgres $PGDATABASE --command='CREATE EXTENSION IF NOT EXISTS hstore;' && \
    osm2pgsql --username=$PGUSER --database=$PGDATABASE --create --cache=10000 --hstore --latlong /data/map.osm.pbf && \
    rm -f /data/map.osm.pbf && \
    pg_ctl stop && \
    echo '# TYPE DATABASE USER ADDRESS METHOD'                > $PGDATA/pg_hba.conf && \
    echo "local all postgres peer"                           >> $PGDATA/pg_hba.conf && \
    echo "local $PGDATABASE $PGUSER           scram-sha-256" >> $PGDATA/pg_hba.conf && \
    echo "host  $PGDATABASE $PGUSER 0.0.0.0/0 scram-sha-256" >> $PGDATA/pg_hba.conf

RUN ls -al $PGDATA

The pipeline yaml(to avoid any value missing, i defined the parameters as variables).

trigger: none

variables:
  - name: tag
    value: $(Build.BuildId)
  - name: DownloadUrls
    value: https://download.geofabrik.de/europe/germany/berlin-latest.osm.pbf
  - name: image_tag
    value: mycontainerregistry.azurecr.io/postgis-test:16-3.4
  - name: POSTGIS_TAG
    value: 16-3.4
  - name: OSM_PASSWORD
    value: testpwd

stages:
- stage: Build
  displayName: Build image
  jobs:
  - job: Build
    displayName: Build
    pool:
      vmImage: Ubuntu-latest
    steps:
    - task: AzureCLI@2
      inputs:
        azureSubscription: 'ARMConn1'
        scriptType: 'bash'
        scriptLocation: 'inlineScript'
        inlineScript: |
          docker system prune --all --force
          docker build --file $(Build.SourcesDirectory)/src/Services/PostGIS/Dockerfile --no-cache --tag $(image_tag) --build-arg POSTGIS_TAG=$(POSTGIS_TAG) --build-arg OSM_PASSWORD=$(OSM_PASSWORD) --build-arg DOWNLOAD_URLS="$(DownloadUrls)" $(Build.SourcesDirectory)/src/

The output in pipeline: