postgresql | create georchestra user fails

[fvanderbiest]

With up to date sources, and no local additions,

TASK: [postgresql | create georchestra user] ********************************** 
failed: [mygeorchestra] => {"failed": true}
msg: unable to connect to database: could not connect to server: No such file or directory
    Is the server running locally and accepting
    connections on Unix domain socket "/var/run/postgresql/.s.PGSQL.5432"?

FATAL: all hosts have already failed -- aborting

PLAY RECAP ******************************************************************** 
           to retry, use: --limit @/home/fvanderbiest/georchestra.retry

mygeorchestra              : ok=17   changed=14   unreachable=0    failed=1   

Ansible failed to complete successfully. Any error output should be
visible above. Please fix these errors and try again.

[landryb]

race condition again i guess, can you retry with this before the 'create georchestra user' task ?

wait_for: port=5432

should be equivalent to

wait_for: path=/var/run/postgresql/.s.PGSQL.5432 state=present

but im not sure having the socket is enough of a clue for psql being fully available.

[fvanderbiest]

Will retry tonight on my home computer. Works without the fix this morning on my c2c computer...

[fvanderbiest]

This did not fix it.

Looks like https://github.com/georchestra/ansible/commit/8efd2304075bb4aed8200f377f7998ecc1211411 is the culprit.

[fvanderbiest]

Seems like a vagrant-specific issue, since this fixes it:

diff --git a/playbooks/georchestra.yml b/playbooks/georchestra.yml
index 9c077d1..f58d3a0 100644
--- a/playbooks/georchestra.yml
+++ b/playbooks/georchestra.yml
@@ -83,7 +83,7 @@
     extractor_datadir: /srv/data/extractor/
     tomcat_keystore_pass: tomcatkstp
     tomcat_basedir: /srv/tomcat
-    system_locale: en_US.UTF-8
+    system_locale: fr_FR.UTF-8
     logs_basedir: /srv/log
     force_https: true
     ldapadmin_adminemail: admin@craig.fr

[landryb]

That.. makes no sense to me :) If you use en_US, postgresql doesnt start in the remote VM ?

[fvanderbiest]

right, makes no sense to me too...

[landryb]

oh, maybe an empty vm settings depends on the host settings, and en_US is not in the list of 'available' locales to locale_gen ?

[fvanderbiest]

possibly

[fvanderbiest]

Correct: http://stackoverflow.com/questions/27292368/vagrantfile-set-locale-for-ssh-agent & https://gist.github.com/mlafeldt/2477731

[landryb]

Eww, gross.. so what's the 'right' way to fix this ? Fix vagrand config/sample invocation ? The playbook ?

[landryb]

http://stackoverflow.com/a/34582489 seems a minimal fix, does it work for you ?

[fvanderbiest]

Will try tonight.

[fvanderbiest]

@Vampouille is going to have a look :-)

[fvanderbiest]

Seems like I fixed this issue by preventing the guest ssh server to accept the LANG and LC_* environment variables from my host:

 diff --git a/Vagrantfile b/Vagrantfile
index fe33e82..3440e1d 100644
--- a/Vagrantfile
+++ b/Vagrantfile
@@ -4,6 +4,12 @@
 # Vagrantfile API/syntax version. Don't touch unless you know what you're doing!
 VAGRANTFILE_API_VERSION = "2"

+$script = <<SCRIPT
+echo configuring ssh in guest...
+sudo sed -i '/AcceptEnv/d' /etc/ssh/sshd_config
+sudo /etc/init.d/ssh restart
+SCRIPT
+
 Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
   # All Vagrant configuration is done here. The most common configuration
   # options are documented and commented below. For a complete reference,
@@ -31,6 +37,8 @@ Vagrant.configure(VAGRANTFILE_API_VERSION) do |config|
   config.vm.define "georchestra" do |georchestra|
   end

+  config.vm.provision "shell", inline: $script
+
   config.vm.provision "ansible" do |ansible|
     # execute this playbook for vm provisioning:
     ansible.playbook = "playbooks/georchestra.yml"

For the record, here are the locales on my host:

LANG=en_US.UTF-8
LANGUAGE=en_US
LC_CTYPE="en_US.UTF-8"
LC_NUMERIC=fr_FR.UTF-8
LC_TIME=fr_FR.UTF-8
LC_COLLATE="en_US.UTF-8"
LC_MONETARY=fr_FR.UTF-8
LC_MESSAGES="en_US.UTF-8"
LC_PAPER=fr_FR.UTF-8
LC_NAME=fr_FR.UTF-8
LC_ADDRESS=fr_FR.UTF-8
LC_TELEPHONE=fr_FR.UTF-8
LC_MEASUREMENT=fr_FR.UTF-8
LC_IDENTIFICATION=fr_FR.UTF-8
LC_ALL=

... while in the box, they are:

LANG=en_US.UTF-8
LANGUAGE=
LC_CTYPE="en_US.UTF-8"
LC_NUMERIC="en_US.UTF-8"
LC_TIME="en_US.UTF-8"
LC_COLLATE="en_US.UTF-8"
LC_MONETARY="en_US.UTF-8"
LC_MESSAGES="en_US.UTF-8"
LC_PAPER="en_US.UTF-8"
LC_NAME="en_US.UTF-8"
LC_ADDRESS="en_US.UTF-8"
LC_TELEPHONE="en_US.UTF-8"
LC_MEASUREMENT="en_US.UTF-8"
LC_IDENTIFICATION="en_US.UTF-8"
LC_ALL=

[fvanderbiest]

An improvement over the previous patch would be:

• run the pre-provisioning step, commenting out the AcceptEnv line
• provision with ansible
• post-provision step, restoring the AcceptEnv line

Opinions ?

[landryb]

http://stackoverflow.com/a/34582489 seems a minimal fix, does it work for you ?

Have you tried this instead ? Seems 'simpler' than a shell hack