integration of update and tile expire procedure

see https://ircama.github.io/osm-carto-tutorials/updating-data/

Implements https://github.com/Overv/openstreetmap-tile-server/issues/2
This commit is contained in:
Steffen Volkmann 2019-06-13 15:39:04 +02:00 committed by Steffen Volkmann
parent d7c2817f9f
commit a9392359b9
4 changed files with 292 additions and 12 deletions

View File

@ -6,6 +6,7 @@ FROM ubuntu:18.04
# Set up environment
ENV TZ=UTC
ENV AUTOVACUUM=on
ENV UPDATES=disabled
RUN ln -snf /usr/share/zoneinfo/$TZ /etc/localtime && echo $TZ > /etc/timezone
# Install dependencies
@ -63,6 +64,10 @@ RUN echo "deb [ allow-insecure=yes ] http://apt.postgresql.org/pub/repos/apt/ bi
unzip \
wget \
zlib1g-dev \
osmosis \
osmium-tool \
cron \
python-psycopg2 python-shapely python-lxml \
&& apt-get clean autoclean \
&& apt-get autoremove --yes \
&& rm -rf /var/lib/{apt,dpkg,cache,log}/
@ -138,7 +143,22 @@ RUN chown -R postgres:postgres /var/lib/postgresql \
&& chown postgres:postgres /etc/postgresql/10/main/postgresql.custom.conf.tmpl \
&& echo "\ninclude 'postgresql.custom.conf'" >> /etc/postgresql/10/main/postgresql.conf
# copy update scripts
COPY openstreetmap-tiles-update-expire /usr/bin/
RUN chmod +x /usr/bin/openstreetmap-tiles-update-expire \
&& mkdir /var/log/tiles \
&& chmod a+rw /var/log/tiles \
&& ln -s /home/renderer/src/mod_tile/osmosis-db_replag /usr/bin/osmosis-db_replag \
&& echo "* * * * * renderer openstreetmap-tiles-update-expire\n" >> /etc/crontab
# install trim_osc.py helper script
USER renderer
RUN cd ~/src \
&& git clone https://github.com/zverik/regional \
&& chmod u+x ~/src/regional/trim_osc.py
# Start running
USER root
COPY run.sh /
ENTRYPOINT ["/run.sh"]
CMD []

View File

@ -9,41 +9,74 @@ First create a Docker volume to hold the PostgreSQL database that will contain t
docker volume create openstreetmap-data
Next, download an .osm.pbf extract from geofabrik.de for the region that you're interested in. You can then start importing it into PostgreSQL by running a container and mounting the file as `/data.osm.pbf`. For example:
docker run -v /absolute/path/to/luxembourg.osm.pbf:/data.osm.pbf -v openstreetmap-data:/var/lib/postgresql/10/main overv/openstreetmap-tile-server import
```
docker run -v /absolute/path/to/luxembourg.osm.pbf:/data.osm.pbf \
-v openstreetmap-data:/var/lib/postgresql/10/main \
overv/openstreetmap-tile-server \
import
```
If the container exits without errors, then your data has been successfully imported and you are now ready to run the tile server.
### optional step to allow consecutive updates of osm extracts
if your import is a extract of planet and is based on polygon then you should download this polynom data and use follwing command for the import procedure:
```
docker run -v /absolute/path/to/luxembourg.osm.pbf:/data.osm.pbf \
-v /absolute/path/to/luxembourg.poly:/data.poly \
-v /openstreetmap-data:/var/lib/postgresql/10/main \
overv/openstreetmap-tile-server \
import
```
## Running the server
Run the server like this:
docker run -p 80:80 -v openstreetmap-data:/var/lib/postgresql/10/main -d overv/openstreetmap-tile-server run
```
docker run -p 80:80 \
-v openstreetmap-data:/var/lib/postgresql/10/main \
-d overv/openstreetmap-tile-server \
run
```
Your tiles will now be available at `http://localhost:80/tile/{z}/{x}/{y}.png`. The demo map in `leaflet-demo.html` will then be available on `http://localhost:80`. Note that it will initially quite a bit of time to render the larger tiles for the first time.
## Preserving rendered tiles
Tiles that have already been rendered will be stored in `/var/lib/mod_tile`. To make sure that this data survives container restarts, you should create another volume for it:
```
docker volume create openstreetmap-rendered-tiles
docker run -p 80:80 -v openstreetmap-data:/var/lib/postgresql/10/main -v openstreetmap-rendered-tiles:/var/lib/mod_tile -d overv/openstreetmap-tile-server run
docker run -p 80:80
-v openstreetmap-data:/var/lib/postgresql/10/main
-v openstreetmap-rendered-tiles:/var/lib/mod_tile
-d overv/openstreetmap-tile-server \
run
```
## Performance tuning and tweaking
## Performance tuning
### update and tile expire procedure
The environment variable "UPDATES" (default value =disabled) allows you to run the container with consecutive updates.
```
docker run -p 80:80 \
-v openstreetmap-data:/var/lib/postgresql/10/main \
-v openstreetmap-rendered-tiles:/var/lib/mod_tile \
-d overv/openstreetmap-tile-server \
-e UPDATES=enabled \
run
```
Details for update procedure and invoked scripts can you find here [link](https://ircama.github.io/osm-carto-tutorials/updating-data/)
### THREADS
The import and tile serving processes use 4 threads by default, but this number can be changed by setting the `THREADS` environment variable. For example:
```
docker run -p 80:80 -e THREADS=24 -v openstreetmap-data:/var/lib/postgresql/10/main -d overv/openstreetmap-tile-server run
```
### AUTOVACUUM
The database use the autovacuum feature by default. This behavior can be changed with `AUTOVACUUM` environment variable. For example:
docker run -p 80:80 -e AUTOVACUUM=off -v openstreetmap-data:/var/lib/postgresql/10/main -d overv/openstreetmap-tile-server
## Benchmarks
### Benchmarks
You can find an example of the import performance to expect with this image on the [OpenStreetMap wiki](https://wiki.openstreetmap.org/wiki/Osm2pgsql/benchmarks#debian_9_.2F_openstreetmap-tile-server).

View File

@ -0,0 +1,207 @@
#!/bin/sh
set -e
#------------------------------------------------------------------------------
# AJT - change directory to mod_tile directory so that we can run replag
# and other things directly from this script when run from cron.
# Change the actual location to wherever installed locally.
#------------------------------------------------------------------------------
export PATH=.:/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin
ACCOUNT=renderer
cd /home/$ACCOUNT/src/mod_tile/
#------------------------------------------------------------------------------
# Extra OSM2PGSQL_OPTIONS may need setting if a tag transform script is
# in use. See https://github.com/SomeoneElseOSM/SomeoneElse-style and
# http://wiki.openstreetmap.org/wiki/User:SomeoneElse/Ubuntu_1404_tileserver_load
# The database name always needs setting.
#------------------------------------------------------------------------------
OSMOSIS_BIN=osmosis
OSM2PGSQL_BIN=osm2pgsql
TRIM_BIN=/home/$ACCOUNT/src/regional/trim_osc.py
DBNAME=gis
OSM2PGSQL_OPTIONS="-d $DBNAME -G --hstore --tag-transform-script /home/renderer/src/openstreetmap-carto/openstreetmap-carto.lua -C 2048 --number-processes ${THREADS:-4} -S /home/renderer/src/openstreetmap-carto/openstreetmap-carto.style"
#------------------------------------------------------------------------------
# When using trim_osc.py we can define either a bounding box (such as this
# example for England and Wales) or a polygon.
# See https://github.com/zverik/regional .
# This area will usually correspond to the data originally loaded.
#------------------------------------------------------------------------------
TRIM_POLY_FILE="/var/lib/mod_tile/data.poly"
TRIM_OPTIONS="-d $DBNAME"
#TRIM_REGION_OPTIONS="-b -14.17 48.85 2.12 61.27"
TRIM_REGION_OPTIONS="-p $TRIM_POLY_FILE"
BASE_DIR=/var/lib/mod_tile
LOG_DIR=/var/log/tiles/
WORKOSM_DIR=$BASE_DIR/.osmosis
LOCK_FILE=/tmp/openstreetmap-update-expire-lock.txt
CHANGE_FILE=$BASE_DIR/changes.osc.gz
EXPIRY_FILE=$BASE_DIR/dirty_tiles
STOP_FILE=$BASE_DIR/stop.txt
OSMOSISLOG=$LOG_DIR/osmosis.log
PGSQLLOG=$LOG_DIR/osm2pgsql.log
EXPIRYLOG=$LOG_DIR/expiry.log
RUNLOG=$LOG_DIR/run.log
#------------------------------------------------------------------------------
# The tile expiry section below can re-render, delete or dirty expired tiles.
# By default, tiles between EXPIRY_MINZOOM and EXPIRY_MAXZOOM are rerendered.
# "render_expired" can optionally delete (and/or dirty) tiles above a certail
# threshold rather than rendering them.
# Here we expire (but don't immediately rerender) tiles between zoom levels
# 13 and 18 and delete between 19 and 20.
#------------------------------------------------------------------------------
EXPIRY_MINZOOM=13
EXPIRY_TOUCHFROM=13
EXPIRY_DELETEFROM=19
EXPIRY_MAXZOOM=20
#*************************************************************************
#*************************************************************************
m_info()
{
echo "[`date +"%Y-%m-%d %H:%M:%S"`] $$ $1" >> "$RUNLOG"
}
m_error()
{
echo "[`date +"%Y-%m-%d %H:%M:%S"`] $$ [error] $1" >> "$RUNLOG"
m_info "resetting state"
/bin/cp $WORKOSM_DIR/last.state.txt $WORKOSM_DIR/state.txt || true
rm "$CHANGE_FILE" || true
rm "$EXPIRY_FILE.$$" || true
rm "$LOCK_FILE"
exit
}
m_ok()
{
echo "[`date +"%Y-%m-%d %H:%M:%S"`] $$ $1" >> "$RUNLOG"
}
getlock()
{
if [ -s $1 ]; then
if [ "$(ps -p `cat $1` | wc -l)" -gt 1 ]; then
return 1 #false
fi
fi
echo $$ >"$1"
return 0 #true
}
freelock()
{
rm "$1"
rm "$CHANGE_FILE"
}
if [ $# -eq 1 ] ; then
m_info "Initialising Osmosis replication system to $1"
mkdir $WORKOSM_DIR
$OSMOSIS_BIN --read-replication-interval-init workingDirectory=$WORKOSM_DIR 1>&2 2> "$OSMOSISLOG"
wget "http://replicate-sequences.osm.mazdermind.de/?"$1"T00:00:00Z" -O $WORKOSM_DIR/state.txt
mv $WORKOSM_DIR/configuration.txt $WORKOSM_DIR/configuration_orig.txt
sed "s!baseUrl=http://planet.openstreetmap.org/replication/minute!baseUrl=https://planet.openstreetmap.org/replication/minute!" $WORKOSM_DIR/configuration_orig.txt > $WORKOSM_DIR/configuration.txt
else
# make sure the lockfile is removed when we exit and then claim it
if ! getlock "$LOCK_FILE"; then
m_info "pid `cat $LOCK_FILE` still running"
exit 3
fi
if [ -e $STOP_FILE ]; then
m_info "stopped"
exit 2
fi
# -----------------------------------------------------------------------------
# Add disk space check from https://github.com/zverik/regional
# -----------------------------------------------------------------------------
MIN_DISK_SPACE_MB=500
if `python -c "import os, sys; st=os.statvfs('$BASE_DIR'); sys.exit(1 if st.f_bavail*st.f_frsize/1024/1024 > $MIN_DISK_SPACE_MB else 0)"`; then
m_info "there is less than $MIN_DISK_SPACE_MB MB left"
exit 4
fi
seq=`cat $WORKOSM_DIR/state.txt | grep sequenceNumber | cut -d= -f2`
m_ok "start import from seq-nr $seq, replag is `osmosis-db_replag -h`"
/bin/cp $WORKOSM_DIR/state.txt $WORKOSM_DIR/last.state.txt
m_ok "downloading diff"
if ! $OSMOSIS_BIN --read-replication-interval workingDirectory=$WORKOSM_DIR --simplify-change --write-xml-change $CHANGE_FILE 1>&2 2> "$OSMOSISLOG"; then
m_error "Osmosis error"
fi
if [ -f TRIM_POLY_FILE ] ; then
m_ok "filtering diff"
/var/lib/mod_tile/data.poly
if ! $TRIM_BIN $TRIM_OPTIONS $TRIM_REGION_OPTIONS -z $CHANGE_FILE $CHANGE_FILE 1>&2 2>> "$RUNLOG"; then
m_error "Trim_osc error"
fi
else
m_ok "filtering diff skipped"
fi
m_ok "importing diff"
#------------------------------------------------------------------------------
# Previously openstreetmap-tiles-update-expire tried to dirty layer
# "$EXPIRY_MAXZOOM - 3" (which was 15) only. Instead we write all expired
# tiles in range to the list (note the "-" rather than ":" in the "-e"
# parameter).
#------------------------------------------------------------------------------
if ! $OSM2PGSQL_BIN -a --slim -e$EXPIRY_MINZOOM-$EXPIRY_MAXZOOM $OSM2PGSQL_OPTIONS -o "$EXPIRY_FILE.$$" $CHANGE_FILE 1>&2 2> "$PGSQLLOG"; then
m_error "osm2pgsql error"
fi
#------------------------------------------------------------------------------
# The lockfile is normally removed before we expire tiles because that is
# something thatcan be done in parallel with further processing. In order to
# avoid rework, if actually rerendering is done rather than just deleting or
# dirtying, it makes sense to move it lower down.
#------------------------------------------------------------------------------
# m_ok "Import complete; removing lock file"
# freelock "$LOCK_FILE"
m_ok "expiring tiles"
#------------------------------------------------------------------------------
# When expiring tiles we need to define the style sheet if it's not "default".
# In this case it's "ajt".
# Previously all tiles on the "dirty" list between $EXPIRY_MINZOOM and
# $EXPIRY_MAXZOOM were dirtied. We currently re-render
# tiles >= $EXPIRY_MINZOOM and < $EXPIRY_DELETEFROM, expiry from 14 and
# delete >= $EXPIRY_DELETEFROM and <= $EXPIRY_MAXZOOM.
# The default path to renderd.sock is fixed.
#------------------------------------------------------------------------------
if ! render_expired --map=ajt --min-zoom=$EXPIRY_MINZOOM --touch-from=$EXPIRY_TOUCHFROM --delete-from=$EXPIRY_DELETEFROM --max-zoom=$EXPIRY_MAXZOOM -s /var/run/renderd/renderd.sock < "$EXPIRY_FILE.$$" 2>&1 | tail -8 >> "$EXPIRYLOG"; then
m_info "Expiry failed"
fi
rm "$EXPIRY_FILE.$$"
#------------------------------------------------------------------------------
# Only remove the lock file after expiry (if system is slow we want to delay
# the next import, not have multiple render_expired processes running)
#------------------------------------------------------------------------------
freelock "$LOCK_FILE"
m_ok "Done with import"
fi

20
run.sh
View File

@ -16,6 +16,7 @@ if [ "$#" -ne 1 ]; then
echo " run: Runs Apache and renderd to serve tiles at /tile/{z}/{x}/{y}.png"
echo "environment variables:"
echo " THREADS: defines number of threads used for importing / tile rendering"
echo " UPDATES: consecutive updates (enabled/disabled)"
exit 1
fi
@ -34,6 +35,20 @@ if [ "$1" = "import" ]; then
if [ ! -f /data.osm.pbf ]; then
echo "WARNING: No import file at /data.osm.pbf, so importing Luxembourg as example..."
wget -nv http://download.geofabrik.de/europe/luxembourg-latest.osm.pbf -O /data.osm.pbf
wget -nv http://download.geofabrik.de/europe/luxembourg.poly -O /data.poly
fi
# determine and set osmosis_replication_timestamp (for consecutive updates)
osmium fileinfo /data.osm.pbf > /var/lib/mod_tile/data.osm.pbf.info
osmium fileinfo /data.osm.pbf | grep 'osmosis_replication_timestamp=' | cut -b35-44 > /var/lib/mod_tile/replication_timestamp.txt
REPLICATION_TIMESTAMP=$(cat /var/lib/mod_tile/replication_timestamp.txt)
# initial setup of osmosis workspace (for consecutive updates)
sudo -u renderer openstreetmap-tiles-update-expire $REPLICATION_TIMESTAMP
# copy polygon file if available
if [ -f /data.poly ]; then
sudo -u renderer cp /data.poly /var/lib/mod_tile/data.poly
fi
# Import data
@ -52,6 +67,11 @@ if [ "$1" = "run" ]; then
# Configure renderd threads
sed -i -E "s/num_threads=[0-9]+/num_threads=${THREADS:-4}/g" /usr/local/etc/renderd.conf
# start cron job to trigger consecutive updates
if [ "$UPDATES" = "enabled" ]; then
/etc/init.d/cron start
fi
# Run
sudo -u renderer renderd -f -c /usr/local/etc/renderd.conf
service postgresql stop