open-future-belgium / postgresql Goto Github PK

View Code? Open in Web Editor NEW

11.0 11.0 30.0 84 KB

Ansible PostgreSQL

License: MIT License

postgresql's People

Contributors

Stargazers

Watchers

postgresql's Issues

sometimes task installing dependencies fails

TASK: [patrik.uytterhoeven.PostgreSQL-For-RHEL6x | Install | Add PostgreSQL repository] *** 

changed: [host27.example.org]

TASK: [patrik.uytterhoeven.PostgreSQL-For-RHEL6x | Install | Make sure CentOS & RHEL dependencies are installed] *** 
failed: [host27.example.org] => {"changed": false, "failed": true, "rc": 0, "results": []}
msg: No Package matching 'python-psycopg2' found available, installed or updated

FATAL: all hosts have already failed -- aborting

can we add update_cache by default?

Newest version of CentOS breaks ansible's modules for postgres user/db modification

I updated to a newer version of CentOS 7 and also the latest stable version of Ansible and suddenly this playbook stopped working for creating users/databases.

Eventually I tracked it down to this playbook sets Postgres to listen on Unix socket /tmp/.s.PGSQL.5432 but the Ansible modules that handle Postgres user/db creation default to /var/run/postgresql/.s.PGSQL.5432

I haven't dug into it to check if Ansible changed where it looks, or if it gets the info from an underlying OS config which might have changed when I updated my CentOS 7 image.

The simple change is to enable listening on Unix socket /var/run/postgresql/.s.PGSQL.5432 by default.

Why is libselinux a dependency for this role?

https://github.com/Open-Future-Belgium/PostgreSQL/blob/master/tasks/RedHat.yml#L3

Why is libselinux required for installing/managing postgres?

I don't see it listed as a dependency anywhere in the ansible postgres modules docs nor do I see it referenced elsewhere in this role.

[feature] Avoid unnecessary restart when postgresql.conf changes

Most postgresql.conf changes don't need a restart, just a reload. And a reload is much safer.

To avoid a restart except where essential, there are a few approaches.
One is documented at http://www.postgresql.org/message-id/[email protected]

My thought was to have postgresql.conf only include postgresql_needs_restart.conf and postgresql_needs_reload.conf. postgresql_needs_restart.conf would include the parameters mentioned in the above email thread.

If postgresql.conf or postgresql_needs_restart.conf changes, the restart postgresql handler is fired as now. But if postgresql_needs_reload.conf changes, a new reload postgresql handler is fired.

I'm happy to implement this but thought I'd see what your thoughts were first on whether it's a sensible approach or if there's anything I've missed.

No longer compatible with ansible 1.9.X

I recently started getting ERROR: block is not a legal parameter in an Ansible task or handler. It looks like a recent commit added the block parameter which only works with ansible 2.0. I'm just checking to see if this was a conscious change to drop support for earlier ansible versions, and if so, to request that the docs be updated to say that. Thanks!

role is not idempotent for CentOS7

task

- name: restart tmpfiles service
  command: systemd-tmpfiles --create
  when: ansible_distribution_major_version == '7'

reports "changed" every ansible run

Not compatible with postgres < 9.3

The config template has parameters that were introduced in 9.3:

LOG:  unrecognized configuration parameter "unix_socket_directories" in file "/var/lib/pgsql/9.2/data/postgresql.conf" line 59
LOG:  unrecognized configuration parameter "wal_sender_timeout" in file "/var/lib/pgsql/9.2/data/postgresql.conf" line 183
LOG:  unrecognized configuration parameter "wal_receiver_timeout" in file "/var/lib/pgsql/9.2/data/postgresql.conf" line 200
LOG:  unrecognized configuration parameter "lock_timeout" in file "/var/lib/pgsql/9.2/data/postgresql.conf" line 360
FATAL:  configuration file "/var/lib/pgsql/9.2/data/postgresql.conf" contains errors

Not working with postgresql 9.5

Works pretty fine with psql 9.4 on CentOS 7.2, but not with 9.5

Following error while running the playbook:

RUNNING HANDLER [patrik.uytterhoeven.PostgreSQL-For-RHEL6x : restart postgresql] ***
fatal: [localhost]: FAILED! => {"changed": false, "failed": true, "msg": "Job for postgresql-9.5.service failed because the control process exited with error code. See \"systemctl status postgresql-9.5.service\" and \"journalctl -xe\" for details.\n"}
        to retry, use: --limit @playbooks/database/main.retry

PLAY RECAP *********************************************************************
localhost                  : ok=23   changed=15   unreachable=0    failed=1

systemctl status postgresql-9.5.service
● postgresql-9.5.service - PostgreSQL 9.5 database server
   Loaded: loaded (/etc/systemd/system/postgresql-9.5.service; disabled; vendor preset: disabled)
   Active: failed (Result: exit-code) since Wed 2016-04-13 07:51:19 CEST; 14s ago
  Process: 6342 ExecStart=/usr/pgsql-9.5/bin/pg_ctl start -D ${PGDATA} -s -w -t 300 (code=exited, status=1/FAILURE)
  Process: 6336 ExecStartPre=/usr/pgsql-9.5/bin/postgresql95-check-db-dir ${PGDATA} (code=exited, status=0/SUCCESS)

Apr 13 07:51:18 nlihmoodle01.hosting.local systemd[1]: Starting PostgreSQL 9.5 database server...
Apr 13 07:51:18 nlihmoodle01.hosting.local pg_ctl[6342]: **LOG:  unrecognized configuration parameter "checkpoint_segments" in file "/var/lib/pgsql/9.5/data/postgresql.conf" line 161**
Apr 13 07:51:18 nlihmoodle01.hosting.local pg_ctl[6342]: **FATAL:  configuration file "/var/lib/pgsql/9.5/data/postgresql.conf" contains errors**
Apr 13 07:51:19 nlihmoodle01.hosting.local pg_ctl[6342]: pg_ctl: could not start server
Apr 13 07:51:19 nlihmoodle01.hosting.local pg_ctl[6342]: Examine the log output.
Apr 13 07:51:19 nlihmoodle01.hosting.local systemd[1]: postgresql-9.5.service: control process exited, code=exited status=1
Apr 13 07:51:19 nlihmoodle01.hosting.local systemd[1]: Failed to start PostgreSQL 9.5 database server.
Apr 13 07:51:19 nlihmoodle01.hosting.local systemd[1]: Unit postgresql-9.5.service entered failed state.
Apr 13 07:51:19 nlihmoodle01.hosting.local systemd[1]: postgresql-9.5.service failed.

Playbook:

  roles:
    - patrik.uytterhoeven.PostgreSQL-For-RHEL6x
  vars:
      postgresql_version: 9.5
      postgresql_admin_user: "postgres"
      postgresql_default_auth_method: "trust"

role is broken for CentOS 6 and postgresql93-server

NOTIFIED: [patrik.uytterhoeven.PostgreSQL-For-RHEL6x | restart postgresql] **** 
failed: [192.168.1.17] => {"failed": true}
msg: Stopping postgresql-9.3 service: [  OK  ]
Starting postgresql-9.3 service: [FAILED]

# grep FATAL pgstartup.log 
2015-07-18 21:40:00 UTC FATAL:  could not create lock file "/var/run/postgresql/.s.PGSQL.5432.lock": No such file or directory

because template renders

# grep -i unix_socket postgresql.conf 
unix_socket_directories = '/tmp,/var/run/postgresql/'

and there's no such dir

# ll /var/run/postgresql
ls: cannot access /var/run/postgresql: No such file or directory

I just can't take ansible seriously because of that stuff.
Seems nobody cares at all about testing and such.
Every single role that I tried to use from galaxy had fatal problems. Sigh.

Yes, I can just to redefine postgresql_unix_socket_directories but come on - isn't it suppose to work by default?

I would suggest you to use test-kitchen or something.

yet another CentOS6 bug with 9.4

task

- name: Initialise Postgresql Database if it is not
  command: service postgresql{{ join_char | default("-")}}{{ postgresql_version }} initdb {{postgresql_cluster_name}}

silently fails for 9.4 with message

cat /var/lib/pgsql/9.4/pgstartup.log 
The files belonging to this database system will be owned by user "postgres".
This user must also own the server process.

initdb: invalid locale name "data"

but it works with 9.3

with every ansible run postgresql restarts

Are those

# {{ ansible_managed }}

really necessary ?
They cause PostgreSQL restarts every time.

postgresql_env

In defaults there's

postgresql_env:
  LC_ALL: "{{ postgresql_locale }}"
  LC_LCTYPE: "{{ postgresql_locale }}"

what is it for?
I've failed to find any mentions of it elsewhere in role.

- name: Initialise Postgresql Database if it is not
  command: service postgresql{{ join_char | default("-")}}{{ postgresql_version }} initdb {{postgresql_cluster_name}}
  when: data_dir.stat.exists == false and ansible_distribution_major_version == '6'

produces this warning in /var/lib/pgsql/9.3/pgstartup.log

initdb: invalid locale name "data"
initdb: invalid locale name "data"
initdb: invalid locale name "data"
initdb: invalid locale name "data"
initdb: invalid locale name "data"
initdb: invalid locale name "data"

plz put the ansible role name in the part of your documentation that is called Role Name:

Exclude postgresql* packages from distribution' repo files

Distribution's repo files should contain a line exclude=postgresql* so that dependencies resolve to proper repository. Source: https://wiki.postgresql.org/wiki/YUM_Installation

open-future-belgium / postgresql Goto Github PK

postgresql's People

Contributors

Stargazers

Watchers

Forkers

postgresql's Issues

Recommend Projects

Recommend Topics

Recommend Org