Page Comparison

Preamble

• Notify all stake holders before the release.
• Before proceeding with initializations, make sure that you are pointing at branches/1.1.3.
• Execute all steps in the SF ENVs first and test the application.
• If tests pass, then execute steps on SD ENVs.
• Should things go badly, we can switch over to SD.

DOWN

web into MAINT mode (SF_PROD_WEB)

put the web servers into maintenance mode (SF WEB, SD WEB)

cd /var/www/html
sudo ./set_eas_mode.sh MAINT

database backup (SF_PROD_DB)

sudo -u postgres -i
/home/dba/scripts/dbbackup.sh > /var/tmp/dbbackup.log

...

DEPLOY

ETL PROD

upgrade ETL code

Log into to the ETL server and do he the following.

REM see warning above
cd "C:\Program Files\MAD\"
del etl
setup.bat

migrate DB

Now you can run this migrate job:

cd etl\src
python job.py --job migrate_to_1.1.3 --action INIT --env SF_PROD

reconfigure geowebcache truncate job

Add an scheduled task to truncate the geowebcache.
The ETL job name is "map_cache_truncate".
This makes obsolete the linux crontab entry on GEO PROD (SF for sure, maybe SD) that truncate the geowebcache...
so you must remove these crontab entires. Execute the line below on sf_prod_geo to being editing the crontab... 

sudo crontab -e

upgrade java (SF_PROD_GEO

...

, SD_PROD_GEO)

install the latest version of java

Remember that in PROD you must also do this to the SD instance!

...

wget --no-cookies --header "Cookie: gpw_e24=http%3A%2F%2Fwww.oracle.com" "http://download.oracle.com/otn-pub/java/jdk/7/jre-7-linux-x64.rpm"
sudo rpm -ivh jre-7-linux-x64.rpm?AuthParam=1347904319_53c7d4517bc08e457c15c859a1ba216e

install JAI

http://download.java.net/media/jai/builds/release/1_1_3/

wget http://download.java.net/media/jai/builds/release/1_1_3/jai-1_1_3-lib-linux-amd64-jre.bin
sudo cp jai-1_1_3-lib-linux-amd64-jre.bin /usr/java/jre1.7.0
sudo cd /usr/java/jre1.7.0
sudo sh jai-1_1_3-lib-linux-amd64-jre.bin

remove previous version of java

cd /usr
sudo rm -rf jdk1.6.0_24

upgrade tomcat (SF_PROD_GEO

...

, SD_PROD_GEO)

preamble

sudo /usr/sbin/usermod -g geodev tomcat
sudo /usr/sbin/usermod -d /usr/share/apache-tomcat-7.0.32 tomcat

install tomcat 7

http://davidghedini.blogspot.com/2011/02/install-tomcat-7-on-centos-rhel.html

wget http://mirror.ccolnevhost.columbia.edunet/pub/software/apache/tomcat/tomcat-7/v7.0.3032/bin/apache-tomcat-7.0.3032.tar.gz
cd /usr/share
sudo tar xvfz ~/apache-tomcat-7.0.3032.tar.gz
sudo chown -Rf tomcat:tomcatgeodev /usr/share/apache-tomcat-7.0.3032/

configure tomcat startup

Ensures that tomcat will startup on server restart.

cd /etc/init.d
sudo svn export http://eas.googlecode.com/svn/trunk/geoserver/tomcat
sudo dos2unix tomcat
sudo chmod 755 tomcat
sudo -u root /sbin/chkconfig --add tomcat
sudo -u root /sbin/chkconfig --level 234 tomcat on
sudo -u root /sbin/chkconfig --list tomcat
sudo chmod g+rwx /usr/share/apache-tomcat-7.0.32/webapps
svn export http://eas.googlecode.com/svn/trunk/geoserver/deploy_geoserver_env.sh /usr/share/apache-tomcat-7.0.32/webapps/deploy_geoserver_env.sh

tomcat start, stop, restart

sudo /sbin/service tomcat start
sudo /sbin/service tomcat stop
sudo /sbin/service tomcat restart

inspect log files for errors

view /usr/share/apache-tomcat-7.0.3032/logs/catalina.out

examine tomcat admin panel

SSH tunnel required.

http://localhost:8081/ (DEV)
http://localhost:8082/ (QA)
http://localhost:8083/ (PROD)

...

sudo rm -rf /usr/share/tomcat5
sudo rm /etc/init.d/tomcat5

upgrade geoserver (SF_PROD_GEO

...

, SD_PROD_GEO)

cd /usr/share/apache-tomcat-7.0.3032/webapps
sudo wget http://downloads.sourceforge.net/geoserver/geoserver-2.1.4-war.zip
sudo unzip geoserver-2.1.4-war.zip
svn
export http://eas.googlecode.com/svn/trunk/geoserver/deploy_geoserver_init.sh
dos2unix deploy_geoserver_init.sh

Clean out the gwc

sudo /sbin/service tomcat stop
sudo rm -rf /srv/gwc
start
sudo /sbin/service tomcat startstop

upgrade the deploy automation and deployClean out the gwc

cd /usr/share/apache-tomcat-7.0.30/webapps
sudo svnrm export-rf http:/srv/eas.googlecode.com/svn/trunk/geoserver/deploy_geoserver_init.sh

sudo ./deploy_geoserver_init.shgwc/*

install geoserver image pyramid plugin

mkdir downloads/
cd downloads/
wget http://iweb.dl.sourceforge.net/project/geoserver/GeoServer%20Extensions/2.1.4/geoserver-2.1.4-pyramid-plugin.zip
unzip geoserver-2.1.4-pyramid-plugin.zip
sudo cp gt-imagepyramid-2.7.5.jar /usr/share/apache-tomcat-7.0.3032/webapps/geoserver/WEB-INF/lib
sudo chown -R tomcat:geodev /usr/share/apache-tomcat-7.0.32/webapps/geoserver
sudo /sbin/service tomcat restartstart

install image processing software (SF_PROD_GEO, SD_PROD_GEO)

We need to install the following on the geoserver boxes so we can process imagery in situ.

Add a python symlink.

install mrsidgeodecode

This software is needed to extract the mrsid data.

mkdir downloads
mkdir bin
cd downloads
wget http://bin.us.lizardtech.com/geoexpress_commandlineutils_linux.tgz
tar xvfz geoexpress_commandlineutils_linux.tgz
cd linux64
tar xvf GeoExpressCLUtils-8.5.0.3330.tar
cp -P GeoExpressCLUtils-8.5.0.3330/bin/* ~/bin

Next, add this to your .bash_profile. Indeed, it is all commented out - we do it because it's the right thing to do.

# If you choose to use lizard tech tools while doing projections, you'll need this.
# NOTE that this prevents GDAL from working properly so I comment it out.
# All I use Lizard tech for is to get data out of mrsid format!
#PROJSO=/home/dev/lizard_tech/linux64/GeoExpressCLUtils-8.5.0.3330/libproject
#export PROJSO

Last, test it to make sure it is working...

• gdal-1.5.4
• geos-3.1.1
• proj-4.6.1
• proj-datumgrid-1.4

web

put SF WEB ..

cd /usr/local/bin/
sudo ln -s /usr/local/bin/python2.5 python 

proj.4

sudo chown dev /usr/local/src
mkdir /usr/local/src/TARFILES
cd /usr/local/src/TARFILES
wget ftp://ftp.remotesensing.org/proj/proj-4.8.0.tar.gz
wget ftp://ftp.remotesensing.org/proj/proj-datumgrid-1.5.tar.gz
cd ../
tar xvfz TARFILES/proj-4.8.0.tar.gz
cd proj-4.8.0/nad
tar xvfz ../../TARFILES/proj-datumgrid-1.5.tar.gz
cd ..
./configure --without-jni
make
sudo make install

install geos

cd /usr/local/src/TARFILES
wget http://download.osgeo.org/geos/geos-3.3.3.tar.bz2
cd ..
tar xvfj TARFILES/geos-3.3.3.tar.bz2
cd geos-3.3.3
# no python bindings
./configure
make
sudo make install

install gdal

cd /usr/local/src/TARFILES
wget http://download.osgeo.org/gdal/gdal-1.9.0.tar.gz
cd ../
tar xvfz TARFILES/gdal-1.9.0.tar.gz
cd gdal-1.9.0/
# set your PATH properly so we get the right version of python
export PATH=/usr/local/bin/:$PATH
./configure --with-python
make
sudo PATH=/usr/local/bin/:$PATH make install 

Add this to your .bashrc

LD_LIBRARY_PATH_DEFAULT=/usr/local/lib
export LD_LIBRARY_PATH_DEFAULT

LD_LIBRARY_PATH=$LD_LIBRARY_PATH_DEFAULT
export LD_LIBRARY_PATH

deploy imagery data (SF_PROD_GEO, SD_PROD_GEO)

setup imagery/geoserver directories

sudo mkdir /srv/gwc
sudo mkdir /srv/imagery
sudo mkdir /srv/imagery/pyramids
sudo mkdir /srv/imagery/pyramids/2012
sudo chown -R tomcat:geodev /srv/imagery/
sudo chown -R tomcat:geodev /srv/gwc/
sudo chmod -R g+rw /srv/imagery/pyramids

copy/prepare the data

You can either build the pyramid or copy the one from SF DEV GEO.
For this deploy, we are going to do the latter.
Do this well before the actual deploy.
Copy

• from SF_DEV_GEO//srv/imagery
• to SF_PROD_GEO//srv/imagery & SD_PROD_GEO//srv/imagery

Use secure FTP which operates over SSH; I use winscp.
This is about 7GB and takes about 15-20 minutes per node.

In the off chance that you need to build the pyramid from scratch, follow the instructions spelled out in main():
http://code.google.com/p/eas/source/browse/trunk/bin/build_pyramids.py

upgrade the deploy automation and deploy (SF_PROD_GEO, SD_PROD_GEO)

cd /usr/share/apache-tomcat-7.0.32/webapps
sudo svn export http://eas.googlecode.com/svn/trunk/geoserver/deploy_geoserver_init.sh
sudo dos2unix deploy_geoserver_init.sh
sudo ./deploy_geoserver_init.sh

After the start up inspect the the log file.

view /usr/share/apache-tomcat-7.0.32/webapps/geoserver_configuration/logs/geoserver.log

This exception is OK because we do not use this plugin:

DEBUG...
Unable to load kakadu JPEG2000 reader spi
java.lang.ClassNotFoundException: it.geosolutions.imageio.plugins.jp2k.JP2KKakaduImageReaderSpi
...

The should be no exceptions besides this one.

deploy web application (SF_PROD_WEB, SD_PROD_WEB)

# see warning at top
cd /var/www/html
sudo ./deploy_eas_init.sh
# make sure web application works and is correct version

UP

web

put WEB (SF/SD) servers into LIVE mode

cd /var/www/html
sudo ./set_eas_mode.sh LIVE

...

and test the application. You will not be able to log into SD.

Put SD WEB servers into STANDBY mode

cd /var/www/html
sudo ./set_eas_mode.sh SD_STANDBY

Seed Image Cache

Once again, log into ETL PROD and seed the image cache:

python job.py --job SEED_IMAGE_CACHE --action EXECUTE --env SF_PROD

This took about 8 hours to run from level 12 to level 19 inclusive.

DONE!