How_I_Setup_Biostar.txt

#############################################################################
#############################################################################
## ALWAYS DO THIS BEFORE YOU WORK ON BIOSTARS:
source ~/virt_env/biostar/bin/activate
## http://simononsoftware.com/virtualenv-tutorial/
## For the initial installs:
source conf/defaults.env
pip install --upgrade -r conf/requirements/base.txt
## Then edit  conf/defaults.env  so that:
## export BIOSTAR_HOSTNAME="0.0.0.0:8080"

#############################################################################
## Then there was the weird filterError that Dan had seen before 
## (from his email)
## Dan followed the instructions in the first answer here:
## http://stackoverflow.com/questions/17474687/npm-install-giving-error-installing-express
## Except that instead of v0.10.12 I picked the most recent version.
## Then he ran:
npm install less
## Setup for this:
    # Install the requirements.
    pip install --upgrade -r conf/requirements/base.txt
    # Load the environment variables.
    source conf/defaults.env
    # Initialize, import test data and run the site.
    ./biostar.sh init import index run
## 'import' will pull in older email files...

## Starting it up:
## This just fires it up with test BS:
## ./biostar.sh index run
##
## Look for it at:   http://gamay:8080
#############################################################################
#############################################################################



#############################################################################
#############################################################################
## ALWAYS DO THIS BEFORE YOU WORK ON BIOSTARS:
source ~/virt_env/biostar/bin/activate

## DB creation and loading:
source ~/virt_env/biostar/bin/activate
source live/loading.env  ## needed to prevent us from spamming everyone while we import emails

## Before you drop the DB, be sure to remove this directory too:
rm -r import/bioc
## THEN to delete, re-create and init postgres DBs
export PGUSER=postgres
./biostar.sh pg_drop pg_create init 
## if the above causes an error saying someone is using the
## database, then restart apache like this:
## sudo service apache2 restart
## THEN import the emails      (R script)
## THEN run the tagging script (R script)


## How to back up and restore:
## To back up, do this
./biostar.sh pg_dump
## You need to enter the postgresql admin password
## This will tell you the name of the file that it is dumping to.
## And before you can restore, you need to drop and recreate
## the database, by doing:
./biostar.sh pg_drop pg_create
## If moving the DB to production, copy the dump file to habu and then ssh to habu
## Then restore with:
./biostar.sh pg_import /home/mcarlson/data/filename
## where filename is the file that was created by pg_dump.

## WHEN you have copied the dump over to Habu (scp
##  /home/mcarlson/data/fullyTagged_biostar-2.2.0-2014-09-05.sql.gz
##  www-data@habu:~/biostar-central/) do it like this:
ssh www-data@habu
workon biostarsenv
cd biostar-central/
source live/custom.env
./biostar.sh pg_drop pg_create
./biostar.sh pg_import NAME_OF_DUMP_FILE_.sql.gz
## Then continue as below with the permissions 


## Then before you can run things you need to set up permissions:
psql -U postgres biostar
## (enter postgres admin password)
grant all privileges on database biostar to biostar;
grant all on all tables in schema public  to biostar;
grant all on all sequences in schema public to biostar;
grant all on all functions in schema public to biostar;
## If on production, also need to do:
update django_site set domain = 'support.bioconductor.org';

## In order to log in as an admin user you need to log into the database
## (psql -U postgres biostar) and then do commands like this (you can do this before dumping)
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'dtenenba@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'mcarlson@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'mtmorgan@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'vobencha@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'sarora@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'nhayden@fhcrc.org';
update users_user set is_admin = TRUE, is_staff = TRUE where email = 'hpages@fhcrc.org';



## Also need to go to admin back and end change ignacio's name to 'Guest User'.
## To do that you need to go here as an admin account:
## https://support.bioconductor.org/admin/


## After restoring, you need to re-enter social login info.
## Unfortunately, even though this information is required to be in the config
## environment that initializes biostar, it also requires it 
## (redundantly) to be in the database as well. So go to

https://support.bioconductor.org/admin/

## and click on Social Apps and then go through each app and fill in the
## appropriate values.

## The canonical place to look for these values is on habu (ssh
## www-data@habu) in the file ~/biostar-central/index.wsgi. This is a
## python script that initializes the app and it's not checked into
## github because it has sensitive information in it.



## Now test it:
## On gamay this means: 
load live/custom.env  ## needed to actually run the site for testing
./biostar.sh run # only ever do this on gamay
## And on production it's just always kept alive.
## and see if you can access http://gamay:8080 or production url


#####################################################################
## older stuff from when we had a sqlite back end:
## Then re-init:
## ALWAYS DO THIS BEFORE YOU WORK ON BIOSTARS:
## source ~/virt_env/biostar/bin/activate
## source live/loading.env
## This deletes all posts from the DB (CAREFUL!)
## ./biostar.sh delete
## ./biostar.sh init
## Then import emails etc. (two R scripts for this)
## THEN index (don't need to run this unless you want to index OLD search boxes)
## This step is only to index the search engine (and we swapped to
##  using a different one of those...
## ./biostar.sh index 
## THEN load new env variables so that the email can work...
source live/custom.env
## Or if you have indexed before, just re-run.
./biostar.sh run
## Then run it
nohup ./biostar.sh init run > biostar.log 2>&1 &

#############################################################################
#############################################################################





#############################################################################
#############################################################################
#############################################################################
## Questions about biostars:
## Can I create an account?  If so, will it still link me to previous
## posts?  Yes, I can create an account, but EVEN IF the emails are
## the same I don't have credit for previous posts...
##
## How can I change the look / tabs?
## Where are the word association file for automatically tagging posts?
## Tagging ANSWER:
## For tagging, I should be able to do it like this:
## http://docs.biostars.org/en/latest/manage.html#command-line-tagging

## How do I set up the automatic email option for users?
## What is difference between 'email' and 'email for every thread'?
## And why is it that every time I make a post I never get any emails?
## I just get some stuff in the server output. Looks like this:

## More questions and answers
## You showed how I can command line give myself a password like this:
## python manage.py usermod -e foo@bar.com -p newpass
## I now know where the DB lives (so I can't ever back it up)
## They are in: 
## ~/tasks/testBiostar/biostar-central/live/ 
## So just go there and 
## sqlite3 biostar.db will get you in...

## test posts (still getting sent to email)
## replies STILL get sent to local stream (instead of out to world)

#####################################################################
## I have to set this up and test.  Instructions are here:
## http://docs.biostars.org/en/latest/customize.html
## looks like I need to just add some python to ./live


#############################################################################
#############################################################################

##Configuring email:
## ad this value to the custom.env file described in this doc
## (http://docs.biostars.org/en/latest/customize.html):
export DJANGO_SETTINGS_MODULE=live.custom

## And this is what goes in the email section of that file

export EMAIL_HOST='mx.fhcrc.org'
export EMAIL_PORT='25'
export EMAIL_HOST_USER='biocbuild@fhcrc.org'
export EMAIL_HOST_PASSWORD=''

## and the DEFAULT_FROM_EMAIL value decides where the email comes from so:
export DEFAULT_FROM_EMAIL='biocbuild@fhcrc.org'

## OK this FINALLY seems to work.  BUT: only if instead of starting with:
source conf/defaults.env
./biostar.sh run

## I instead have to start with:
source live/custom.env
./biostar.sh run

########################################################################
########################################################################



########################################################################
########################################################################
## Skinning:
## art is in: 
## /home/mcarlson/tasks/testBiostar/biostar-central/biostar/static
## biostar2.logo.png
## And this is referred to here (pointed this to new image)
## biostar/settings/base.py:SITE_LOGO = "biostar2.logo.png"
## This above change was formalized by adding the following to live/custom.py
## # The site logo.
## SITE_LOGO = "bioconductor_logo_rgb_small.jpg"
##
## style stuff is in:
## biostar/static/biostar.style.less
##
## Colors:
## #5b8c3e: forest green
## #6b3e4a: DARK maroon
## #E9530F: orange?
## #8f2c47: unanswered color default (cranberry)
##
## And in fact some colors have not yet been factored out of here:
## biostar/static/biostar.base.less 
## (They still need to be moved over to here:)
## biostar/static/biostar.style.less
## This was done before and submitted to the core
##
########################################################################
## Bioc Colors  (as grepped from our web site)
## BiocGreen: #87b13f
## BiocBlue: #1881c2
## And our logos:
## http://www.bioconductor.org/about/logo/
##
########################################################################
########################################################################



#########################################################################
## Dans emails about the production server:
## Everything is hosted on habu. I *think* you should be able to
## ssh mcarlson@habu
## without needing a password. Let me know if that works.
## And everything runs as the www-data user (this is the user that
## apache runs as). So you should not make any changes as mcarlson,
## instead you should:
## sudo su - www-data
## 
## AMMENDMENT: now just ssh in directly as the www-data user:

ssh www-data@habu

## You should have sudo privileges so that should work, let me know if not.

## I tried to set up ssh'ing directly to www-data@habu and for
## whatever reason it doesn't work. I can look into it but this is a
## not too annoying workaround.

## Anyway, inside ~www-data you'll see a biostar-central
## directory. This is a clone of our fork of biostars.

## The way that Apache links to biostar is through the index.wsgi file
## inside the biostar-central directory, which is a python script.

## Since apache calls this script directly, there is no opportunity to
## source a .sh file like you do on gamay, so all the environment
## variables that were set in live/custom.env are now set in
## index.wsgi. They are just set in python instead of shell script.

## One more thing, biostar runs in a virtualenv on habu just like it
## does on gamay, and you can activate that virtualenv as follows:

source ~/.virtualenvs/biostarsenv/bin/activate

## ALSO: on production there should never be a need to run the "run" script
## So no: ./biostar.sh run
## ALSO: do NOT ever load this on production since the envir there is
## not able to be set to 'safe' - IOW the always active index.wsgi
## apache file is based only on the unsafe contents of live/custom.env
## AND: if you must restart apache you do it like this:

sudo service apache restart

## once things are finished indexing on gamay, I need to 1) copy the
## contents of /live/custom.py from gamay 2) re-run the init script,
## 3) copy the DB, and then apache should be running it.

## Changing default biostar templates
## The templates are in biostar/server/templates

## Dan changed starbase.html in this directory to make
## it very clear that the site is not live yet. There
## are comments indicating what changed. This content
## should be removed when we go live.



################################################################################
## Notes from Dan on how to best merge in 'improvements' from the istvan branch.
## 
## Hi Marc,

## I went ahead and pulled and merged this. Here is how I did it, for
## your future reference:

## cd tasks/forkBiostars/biostar-central/
## git branch # show what branch you're on
## git checkout istvan # fails because some stuff is not committed
## # so commit it:
## git commit -a
## # now try again
## git checkout istvan # now works
## git pull istvan
## git checkout master # switch back to master
## git merge istvan
## # hopefully the merge will succeed w/o conflicts. If not, let me know.
## git status
## git push # enter your github username and password when prompted