Re: [pve-devel] [PATCH pve-storage 3/8] cephconfig: support sections in the format of [client.$NAME]

["Fabian Grünbichler" <f.gruenbichler@proxmox.com>]

On January 30, 2024 7:40 pm, Max Carrara wrote:
> Signed-off-by: Max Carrara <m.carrara@proxmox.com>
> ---
>  src/PVE/CephConfig.pm | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/src/PVE/CephConfig.pm b/src/PVE/CephConfig.pm
> index 6b10d46..46b92ea 100644
> --- a/src/PVE/CephConfig.pm
> +++ b/src/PVE/CephConfig.pm
> @@ -77,6 +77,7 @@ sub write_ceph_config {
>  
>      &$cond_write_sec('global');
>      &$cond_write_sec('client');
> +    &$cond_write_sec('client\..*');
>  
>      &$cond_write_sec('mds');
>      &$cond_write_sec('mon');

this whole code is a bit weird (pre-existing, not your patch in
particular)..

should we maybe switch it to
- keep track of sections which were already written
- write out all not-yet-written sections as a last step?

else, a RMW cycle might lose config sections just because this code is
not aware of them? while we're at it, double-checking how the ceph
parser handles sections with whitespace in their name and other funny
business might be a good idea, just to prevent any discrepancy between
our parser and theirs..