From 18d68ae22e27186ff84a36be8869ad5dcd89ef0c Mon Sep 17 00:00:00 2001
From: Stefan Verhoeven <stefan.verhoeven@gmail.com>
Date: Sun, 23 Feb 2020 17:43:39 +0100
Subject: [PATCH] Recompile scipy and sklearn to prevent numpy importerror

---
 Dockerfile            | 7 +++++--
 create_dev_user       | 3 +++
 setup_organization.sh | 2 ++
 3 files changed, 10 insertions(+), 2 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index fa63fc1..ecee846 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -66,6 +66,7 @@ RUN useradd -m -d /home/cartodb -s /bin/bash cartodb && \
     postgresql-12-plproxy \
     postgresql-12-postgis-3 \
     postgresql-12-postgis-3-scripts \
+    postgresql-12-mysql-fdw \
     postgis \
     liblwgeom-2.4-0 \
     ca-certificates \
@@ -115,6 +116,7 @@ RUN useradd -m -d /home/cartodb -s /bin/bash cartodb && \
     liblapack-dev \
     gfortran \
   --no-install-recommends && \
+  apt-get upgrade -y -q && \
   rm -rf /var/lib/apt/lists/*
 
 RUN git config --global user.email you@example.com
@@ -166,8 +168,9 @@ RUN cd / && \
     pip3 install --no-cache-dir scipy && \
     make install && \
     # Numpy gets upgraded after scikit-learn is installed
+    # `import sklearn` fails with `RuntimeError: module compiled against API version 0xb but this version of numpy is 0xa` and `ImportError: numpy.core.multiarray failed to import`
     # make sure scikit-learn is compatible with currently installed numpy, by reinstalling
-    pip3 install --force-reinstall --no-cache-dir scikit-learn==0.17.0 && \
+    pip3 install --force-reinstall --no-cache-dir scipy==0.17.0 scikit-learn==0.17.0 && \
     cd ..
 
 # Initialize template postgis db
@@ -196,7 +199,7 @@ RUN git clone --recursive git://github.com/CartoDB/cartodb.git && \
     git checkout $CARTODB_VERSION && \
     # Install cartodb extension
     cd lib/sql && \
-    PGUSER=postgres make install && \
+    PGUSER=postgres make all install && \
     service postgresql start && /bin/su postgres -c \
       /tmp/cartodb_pgsql.sh && service postgresql stop && \
     cd - && \
diff --git a/create_dev_user b/create_dev_user
index 411a56d..04d656a 100755
--- a/create_dev_user
+++ b/create_dev_user
@@ -36,3 +36,6 @@ bundle exec  rake cartodb:db:configure_geocoder_extension_for_non_org_users[$SUB
 
 # Enable sync tables
 echo "UPDATE users SET sync_tables_enabled=true WHERE username='${SUBDOMAIN}'" | psql -U postgres -t carto_db_development
+
+# Set import limit to 200Mb and 4M rows
+bundle exec rake cartodb:set_custom_limits_for_user[$SUBDOMAIN,209715200,4000000,5]
diff --git a/setup_organization.sh b/setup_organization.sh
index 849b12b..8d1b30d 100644
--- a/setup_organization.sh
+++ b/setup_organization.sh
@@ -8,6 +8,8 @@ bundle exec rake cartodb:db:set_unlimited_table_quota["${USERNAME}"]
 bundle exec rake cartodb:db:create_new_organization_with_owner ORGANIZATION_NAME="${ORGANIZATION_NAME}" USERNAME="${USERNAME}" ORGANIZATION_SEATS=100 ORGANIZATION_QUOTA=102400 ORGANIZATION_DISPLAY_NAME="${ORGANIZATION_NAME}"
 bundle exec rake cartodb:db:set_organization_quota[$ORGANIZATION_NAME,5000]
 bundle exec rake cartodb:db:configure_geocoder_extension_for_organizations[$ORGANIZATION_NAME]
+# Set import limit to 200Mb and 4M rows
+bundle exec rake cartodb:set_custom_limits_for_user[$ORGANIZATION_NAME,209715200,4000000,5]
 
 # Enable sync tables
 echo "UPDATE users SET sync_tables_enabled=true WHERE username='${USERNAME}'" | psql -U postgres -t carto_db_development