Arch Linux

Description:

When running a system with python >=3.10 installed, it appears that a macro is not defined
early enough in the upstream's compile, which leads to the standard methods of querying a
ceph cluster's status (ceph -s, ceph health [detail]) to fail with the following message:

'''
HEALTH_ERR Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
[ERR] MGR_MODULE_ERROR: Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
'''

There is a single issue open in the upstream for a related error in the build process[1]
which will likely fix this if/when it's backported to 15.x, but it appears that generally,
the upstream developers do not use or test against python >3.9, which is why I'm opening this
here, as a distro bug rather than against the upstream directly.

Feel free to close it, as _this does not appear to impact ceph itself,_ merely monitoring capacities.
I did feel like the maintainers should be aware of this issue, and did not see it mentioned on the forums
or elsewhere in this bug tracker.

Thanks.

Additional info:
* package version(s)

'''
$ pacman -Qi ceph | head -n2
Name : ceph
Version : 15.2.14-5
'''

* Monitor logs w/ error

'''
2022-01-09T17:50:00+0000 <host> ceph-mon[585]: <time> 7f56c7615640 -1 log_channel(cluster) log [ERR] : overall HEALTH_ERR Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
2022-01-09T18:00:00+0000 <host> ceph-mon[585]: <time> 7f56c7615640 -1 log_channel(cluster) log [ERR] : overall HEALTH_ERR Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
2022-01-09T18:10:00+0000 <host> ceph-mon[585]: <time> 7f56c7615640 -1 log_channel(cluster) log [ERR] : overall HEALTH_ERR Module 'devicehealth' has failed: PY_SSIZE_T_CLEAN macro must be defined for '#' formats
'''

Steps to reproduce:
1. Fresh archlinux install
2. Install ceph == 15.2.14-5
3. Start a local cluster
4. Observe the output of either 'ceph -s' or 'ceph health'

Links:

[1] https://tracker.ceph.com/issues/53441