lvol module: fails on repeated run

[inittab]

This task:
name: Create LVs for web2 DRBD
lvol:
vg=sistema
lv=drbd_web2
size=40G
state=present

Will create the mentioned LV on the first run, and fail in the next runs with:
failed: [patty] => {"err": " New size (10240 extents) matches existing size (10240 extents)\n Run `lvreduce --help' for more information.\n", "failed": true, "item": "", "rc": 3}
msg: Unable to resize drbd_web2 to 40G

Whereas this one, will work without problems:
name: Create LVs for web1 DRBD
lvol:
vg=sistema
lv=drbd_web1
size=7G
state=present

Main difference being the size (and name). Anything above 13-15G will fail. I say 13-15 because I could not always reproduce the issue in that range, but always above it.
Running ansible 1.4 against Debian Wheezy.

[mscherer]

Could you edit the bug so the formatting is right ( mostly using 4 spaces to handle code correctly )

[apollo13]

@mscherer Copied from my config:

- name: mysql volume
  lvol: lv=mysql size=70G vg=main

Can confirm that it fails. Issue is resolved by changing '70G' to '70g'.

[risaacson]

I just tested this against (ansible 1.5 (devel 0a83f74) last updated 2014/02/24 14:32:43 (GMT -500)) and found no issues. I was testing against a recent Debian Wheezy(64-bit) instance creating a 70G LV.

I used the following playbook for testing which should have caught what you saw. Unfortunately up to three months has happened while this ticket has been open and many fixes have come through. Please feel free to retest via devel or a more recent version and reopen this issue if you can confirm that the problem is still there.

---
# This playbook creates and removes logical volumes.

- name: create the lvm lvol - one
  lvol: lv=one size=70g vg=main state=present

- name: remove the lvm lvol - one
  lvol: lv=one vg=main state=absent

- name: create the lvm lvol - two
  lvol: lv=two size=70G vg=main state=present

- name: remove the lvm lvol - two
  lvol: lv=two vg=main state=absent

[risaacson]

In my previous response I was answering a question that I don't think that you have.

The behavior that you are seeing is that when you run for a second time at the same size or larger there is a failure on an existing volume. This is because the second run is trying to resize the volume. You can use a larger(if there is disk space) or a smaller size but the same size will currently fail.

So keeping that in mind the module should check if the lvol exists under the expected vg and if it is the same size to exit out nicely stating that there was no changes.

@inittab let me know if you think that sounds wonky.

[inittab]

Hi, the issue seems to be fixed in 1.4.5. I can run this task over and over without problems (either using g or G as units). I also tried successfully with 1.5 (6fe369f). Closing accordingly. Thanks!

[tom--]

This is not fixed in my system.

• Control: OS X 10.9.3 Ansible 1.6.3
• Target: Ubuntu 14.04

This playbook fails with the output given below:

- hosts: trusty5
  remote_user: "{{ unpriv_user }}"
  sudo: yes
  tasks:
    - lvol: vg=trusty5-vg lv=test size=100G state=present
    - command: lvs --noheadings -o lv_name,size --units G --separator ';' trusty5-vg
      register: lvs
    - debug: var=lvs
    - lvol: vg=trusty5-vg lv=test size=100G state=present

TASK: [lvol vg=trusty5-vg lv=test size=100G state=present] ********************
changed: [trusty5]

TASK: [command lvs --noheadings -o lv_name,size --units G --separator ';' trusty5-vg] ***
changed: [trusty5]

TASK: [debug var=lvs] *********************************************************
ok: [trusty5] => {
    "lvs": {
        "changed": true,
        "cmd": [
            "lvs",
            "--noheadings",
            "-o",
            "lv_name,size",
            "--units",
            "G",
            "--separator",
            ";",
            "trusty5-vg"
        ],
        "delta": "0:00:00.012904",
        "end": "2014-06-23 15:37:29.609964",
        "invocation": {
            "module_args": "lvs --noheadings -o lv_name,size --units G --separator ';' trusty5-vg",
            "module_name": "command"
        },
        "rc": 0,
        "start": "2014-06-23 15:37:29.597060",
        "stderr": "",
        "stdout": "  root;100.00G\n  test;107.37G",
        "stdout_lines": [
            "  root;100.00G",
            "  test;107.37G"
        ]
    }
}

TASK: [lvol vg=trusty5-vg lv=test size=100G state=present] ********************
failed: [trusty5] => {"failed": true}
msg: Sorry, no shrinking of test without force=yes.

There is a discrepancy between lvs(8) and lvcreate(8).

LVS(8)
       --units hHbBsSkKmMgGtTpPeE
              All  sizes  are  output  in these units: (h)uman-readable, (b)ytes, (s)ectors, (k)ilobytes, (m)egabytes, (g)igabytes,
              (t)erabytes, (p)etabytes, (e)xabytes.  Capitalise to use multiples of 1000 (S.I.) instead of 1024.  Can also  specify
              custom units e.g. --units 3M

LVCREATE(8)
       -L, --size LogicalVolumeSize[bBsSkKmMgGtTpPeE]
              Gives the size to allocate for the new logical volume.  A size suffix of K for kilobytes,  M  for  megabytes,  G  for
              gigabytes, T for terabytes, P for petabytes or E for exabytes is optional.
              Default unit is megabytes.

So lvcreate -L 100G is the same as lvcreate -L 100g but lvs --units G is different from lvs --units g.

The lvol module's docs say that it uses sizes as does lvcreate. This is not correct. It creates as does lvcreate but reads current size as does lvs so using capital multipliers will never be idempotent.

A solution would be to lowercase the variable unit passed to lvol on line 156 here.

[jmdh]

This hasn't yet been fixed: discussion on it appears to have moved to: ansible/ansible-modules-extras#196