Cause: `ceph config get` doesn't allow getting value when entity is 'global'
Consequence: it makes the module fail because the module first tries to get the
current value of the option for idempotency concern
Fix: makes the module check in `ceph config dump` instead.
Result: Setting or getting global configuration works.
DescriptionTridibesh Chakraborty
2023-04-20 12:56:34 UTC
Description of problem:
Customer is unable to set/get Global configurations via cephadm-ansible module ceph_config. It is failing complaining that it is unable to get the current value of the global configuration value, hence it is failing.
~~~
fatal: [rhcs5-admin]: FAILED! => changed=false
ansible_facts:
discovered_interpreter_python: /usr/libexec/platform-python
msg: Can't get current value. who=global option=osd_pool_default_size
rc: 1
~~~
Version-Release number of selected component (if applicable):
RHCS : 16.2.10-160.el8cp
Cephadm-ansible: 1.14.0-1.el8cp
cephadm: 16.2.10-160.el8cp
How reproducible:
- It is happening all the time in both customer environment and in my test lab
Steps to Reproduce:
1. Install cephadm and cephadm-ansible
2. Write a playbook with the task using ceph_config module
3. Use global on the `who`section to set/get any global configuration value (e.g. osd_pool_default_size)
4. Run the playbook to set/get the value
5. It will fail complaining below message
~~~
msg: Can't get current value. who=global option=osd_pool_default_size
~~~
6. If I change it from global to any other type e.g. osd or mon, the playbook is running successfully
Actual results:
The ansible playbook is failing to set/get any global configuration value
Expected results:
The playbook should be able to set/get the value
Additional info:
If I run `ceph config set global <option> <value>`, it is successfully setting the value, where as playbook is failing for the same.
~~~
[root@rhcs5-admin cephadm-ansible]# ceph config set global osd_pool_default_size 2
[root@rhcs5-admin cephadm-ansible]# ceph config dump |grep osd_pool_default_size
global advanced osd_pool_default_size 2
[root@rhcs5-admin cephadm-ansible]#
~~~
I have an observation, although the set value is showing in `ceph config dump`, I am unable to get the same via `ceph config get global <option>` command. It is failing throwing the error `Error EINVAL: unrecognized entity 'global'`. It is working fine for other types like mon, osd etc. So I am assuming as the config get command is not working for global configs, the playbook is unable to fetch the current value and failing to set it.
~~~
[root@rhcs5-admin cephadm-ansible]# ceph config get global osd_pool_default_size
Error EINVAL: unrecognized entity 'global'
[root@rhcs5-admin cephadm-ansible]# ceph config get osd osd_pool_default_size
2
[root@rhcs5-admin cephadm-ansible]#
~~~
Also, I have couple of observations from the /usr/share/cephadm-ansible/library/ceph_config.py file
If I see the description of `who` I can see it is pointing to daemons. So if it is true, then I guess we can't set global config values. We need to use `mon`, `mgr`, `osd`, `client` etc, not `global`.
~~~
who:
description:
- which daemon the configuration should be set to
required: true
~~~
Then we need to modify the example section in the file where it gives example of fetching a global config value. Also, in get operation, I believe `value` should not be there. This might be a typo.
~~~
- name: get osd_pool_default_size value
ceph_config:
action: get
who: global
option: osd_pool_default_size
value: 1 <== Might be a typo
~~~
In this case also, customer may still want to set global values via cephadm-ansible playbook.
If the ceph_config module is supposed to set/get global config values, then we need this to be fixed as it is failing.
Please note, I have tested the same on RHCS 6 and it is failing there also.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory (Red Hat Ceph Storage 5.3 Bug Fix update), and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.
https://access.redhat.com/errata/RHBA-2023:3259
Description of problem: Customer is unable to set/get Global configurations via cephadm-ansible module ceph_config. It is failing complaining that it is unable to get the current value of the global configuration value, hence it is failing. ~~~ fatal: [rhcs5-admin]: FAILED! => changed=false ansible_facts: discovered_interpreter_python: /usr/libexec/platform-python msg: Can't get current value. who=global option=osd_pool_default_size rc: 1 ~~~ Version-Release number of selected component (if applicable): RHCS : 16.2.10-160.el8cp Cephadm-ansible: 1.14.0-1.el8cp cephadm: 16.2.10-160.el8cp How reproducible: - It is happening all the time in both customer environment and in my test lab Steps to Reproduce: 1. Install cephadm and cephadm-ansible 2. Write a playbook with the task using ceph_config module 3. Use global on the `who`section to set/get any global configuration value (e.g. osd_pool_default_size) 4. Run the playbook to set/get the value 5. It will fail complaining below message ~~~ msg: Can't get current value. who=global option=osd_pool_default_size ~~~ 6. If I change it from global to any other type e.g. osd or mon, the playbook is running successfully Actual results: The ansible playbook is failing to set/get any global configuration value Expected results: The playbook should be able to set/get the value Additional info: If I run `ceph config set global <option> <value>`, it is successfully setting the value, where as playbook is failing for the same. ~~~ [root@rhcs5-admin cephadm-ansible]# ceph config set global osd_pool_default_size 2 [root@rhcs5-admin cephadm-ansible]# ceph config dump |grep osd_pool_default_size global advanced osd_pool_default_size 2 [root@rhcs5-admin cephadm-ansible]# ~~~ I have an observation, although the set value is showing in `ceph config dump`, I am unable to get the same via `ceph config get global <option>` command. It is failing throwing the error `Error EINVAL: unrecognized entity 'global'`. It is working fine for other types like mon, osd etc. So I am assuming as the config get command is not working for global configs, the playbook is unable to fetch the current value and failing to set it. ~~~ [root@rhcs5-admin cephadm-ansible]# ceph config get global osd_pool_default_size Error EINVAL: unrecognized entity 'global' [root@rhcs5-admin cephadm-ansible]# ceph config get osd osd_pool_default_size 2 [root@rhcs5-admin cephadm-ansible]# ~~~ Also, I have couple of observations from the /usr/share/cephadm-ansible/library/ceph_config.py file If I see the description of `who` I can see it is pointing to daemons. So if it is true, then I guess we can't set global config values. We need to use `mon`, `mgr`, `osd`, `client` etc, not `global`. ~~~ who: description: - which daemon the configuration should be set to required: true ~~~ Then we need to modify the example section in the file where it gives example of fetching a global config value. Also, in get operation, I believe `value` should not be there. This might be a typo. ~~~ - name: get osd_pool_default_size value ceph_config: action: get who: global option: osd_pool_default_size value: 1 <== Might be a typo ~~~ In this case also, customer may still want to set global values via cephadm-ansible playbook. If the ceph_config module is supposed to set/get global config values, then we need this to be fixed as it is failing. Please note, I have tested the same on RHCS 6 and it is failing there also.