The following steps are things that aren’t been handled by Ansible and that were required to do manually so we could run FxA on a Vagrant VM using mozilla/fxa-dev.
Note that this procedure has been made after the steps described in mozilla/fxa-dev#81
-
Inside the VM, create manually the database
sudo -s mysql create database fxa; quit cd /data/fxa-auth-db-server/ -
make sure config/staging.json has root has no password
su app NODE_ENV=stage node bin/db_patcher.js chdir=/data/fxa-auth-db-server exit supervisorctl restart fxa-auth-db-server cd /data/fxa-auth-server/ su app -
as per scripts/start-server.sh, run manually without STDERR
node ./bin/key_server.js | node ./bin/notifier.js -
boomnpm module isn’t found at that version anymore, changepackage.jsonfrom '2.x.x' to '~2.6'vi /data/fxa-auth-server/node_modules/hawk/node_modules/cryptiles/package.json -
Try again
npm update exit mkdir /data/fxa-auth-server/node_modules/fxa-auth-db-mem/node_modules chown app:app /data/fxa-auth-server/node_modules/fxa-auth-db-mem/node_modules su app npm update exit mkdir /data/fxa-auth-server/node_modules/grunt/node_modules/grunt-legacy-log/node_modules/colors chown app:app /data/fxa-auth-server/node_modules/grunt/node_modules/grunt-legacy-log/node_modules/colors -
Complaints about no permission, trying as root now.
npm update -
Module not found "libs/minimal"
cd /data/fxa-auth-server/node_modules/browserid-crypto/libs/ su app ./bundle.sh cd /data/fxa-auth-server node /data/fxa-auth-server/scripts/gen_keys.js node ./bin/key_server.js | node ./bin/notifier.js -
If it doesn’t die, we are good to go
exit supervisorctl start fxa-auth-server
-
OAuth server now
supervisorctl start fxa-oauth-server supervisorctl status -
But it dies
cd ../fxa-oauth-server/ su app -
Run directly without STDERR redirection
CONFIG_FILES="/data/fxa-oauth-server/config/awsbox.json,/data/fxa-oauth-server/config/local.json" node /data/fxa-oauth-server/bin/server.js -
Its a database access problem
vi config/local.json -
Adjust password field. Here is how its configured in ansible config
// From fxa-dev checkout vi roles/oauth/defaults/main.yml // Adjust the `oauth_db_password` field, otherwise python will write "None" and will try to connect to DB using "None" as a password. -
Back in the VM, try again
su app CONFIG_FILES="/data/fxa-oauth-server/config/awsbox.json,/data/fxa-oauth-server/config/local.json" node /data/fxa-oauth-server/bin/server.js -
If it doesn’t die, we’re good to go
exit supervisorctl start fxa-oauth-server
supervisorctl start fxa-rp
Might say its not running. To fix;
cd ../fxa-rp/
cat /etc/supervisor.d/fxa-rp.conf
-
Missing certificate
su app node scripts/gen_keys.js exit supervisorctl start fxa-rp -
Check the logs
tail /var/log/fxa-rp.log // ... Error: Cannot find module 'ejs' -
Missing modules w/ npm modules
suu app npm install -
Dependencies doesn’t have their dependencies handled ^^;;;;;
cd node_modules/connect-fonts/ npm install cd ../../ cd node_modules/connect-fonts/node_modules/node-font-face-generator/ npm install -
Try again
cd ../../../../ CONFIG_FILES="/data/fxa-rp/config.json,/data/fxa-rp/local.json" node /data/fxa-rp/server.js exit -
If fxa-rp doesn’t die, we’re good to go there too.
Note that the Redis error is fine if you are in dev mode though.
supervisorctl start fxa-rp
supervisorctl status
elasticsearch RUNNING pid 17636, uptime 1:18:41
fxa-auth-db-server RUNNING pid 18116, uptime 0:56:41
fxa-auth-mailer FATAL Exited too quickly (process log may have details)
fxa-auth-server RUNNING pid 18872, uptime 0:16:46
fxa-content-server RUNNING pid 17635, uptime 1:18:41
fxa-customs-server RUNNING pid 17631, uptime 1:18:41
fxa-oauth-server RUNNING pid 18987, uptime 0:09:25
fxa-profile:profile-server RUNNING pid 17646, uptime 1:18:41
fxa-profile:profile-static RUNNING pid 17642, uptime 1:18:41
fxa-profile:profile-worker RUNNING pid 17643, uptime 1:18:41
fxa-rp RUNNING pid 19277, uptime 0:00:14
hekad RUNNING pid 17649, uptime 1:18:41
cd ../fxa-auth-mailer
vi /etc/supervisor.d/fxa-auth-mailer.conf
su app
node /data/fxa-auth-mailer/bin/server.js
-
Something about an error w/ Restify
npm install -
Try again, should work now
node /data/fxa-auth-mailer/bin/server.js -
If it doesn’t die, we’re good to go
exit supervisorctl start fxa-auth-mailer
supervisorctl status
elasticsearch RUNNING pid 17636, uptime 1:28:15
fxa-auth-db-server RUNNING pid 18116, uptime 1:06:15
fxa-auth-mailer RUNNING pid 19400, uptime 0:00:06
fxa-auth-server RUNNING pid 18872, uptime 0:26:20
fxa-content-server RUNNING pid 17635, uptime 1:28:15
fxa-customs-server RUNNING pid 17631, uptime 1:28:15
fxa-oauth-server RUNNING pid 18987, uptime 0:18:59
fxa-profile:profile-server RUNNING pid 17646, uptime 1:28:15
fxa-profile:profile-static RUNNING pid 17642, uptime 1:28:15
fxa-profile:profile-worker RUNNING pid 17643, uptime 1:28:15
fxa-rp RUNNING pid 19277, uptime 0:09:48
hekad RUNNING pid 17649, uptime 1:28:15
Now that we have everything running, we need to have it accessible from outside the VM.
iptables -L
sudo iptables -A INPUT -p tcp --dport 22 -j ACCEPT
sudo iptables -A INPUT -p tcp --dport 80 -j ACCEPT
iptables --flush
We’ll have to deal with this later #TODO
-
http://www.if-not-true-then-false.com/2011/install-nginx-php-fpm-on-fedora-centos-red-hat-rhel/
-
yum install policycoreutils-python nmap nmap -p 8080 localhost semanage port --list | grep http_port_t > http_port_t tcp 80, 81, 443, 488, 8008, 8009, 8443, 9000
-
List all ports we need and make sure they are
http_port_t. #TODO -
Enable them in SELinux manually the ones missing
semanage port --add --type http_port_t --proto tcp 9010 semanage port --add --type http_port_t --proto tcp 9011 semanage port --add --type http_port_t --proto tcp 9012 semanage port --add --type http_port_t --proto tcp 7000 semanage port --add --type http_port_t --proto tcp 1113 semanage port --add --type http_port_t --proto tcp 8000 semanage port --add --type http_port_t --proto tcp 3030 semanage port --add --type http_port_t --proto tcp 4900 -
Change port of fxa-oauth-server
vi /data/fxa-oauth-server/config/local.json -
Add a block:
"server": {"port": 9110}, -
Ensure its open
supervisorctl restart fxa-oauth-server semanage port --add --type http_port_t --proto tcp 9110 nmap -p 9110 localhost
vi /etc/nginx/conf.d/upstream/http_fxa_oauth.conf
server 127.0.0.1:9110
FxA content server didnt have npm modules installed
cd /data/fxa-content-server
npm install
su data
bower install
exit
grunt build:dist
chown -R app:app dist/
Make sure we have dist/bower_components with more than 2 folders
supervisorctl restart fxa-content-server