[pve-devel] [PATCH storage] api: fix get content call for volumes

[Fiona Ebner]

Am 06.03.23 um 11:36 schrieb Christian Ebner:
> 
>> On 06.03.2023 11:17 CET Fiona Ebner <f.ebner at proxmox.com> wrote:
>>
>>  
>> Am 06.03.23 um 10:37 schrieb Christian Ebner:
>>> `pvesh get /nodes/{node}/storage/{storage}/content/{volume}` failed for
>>> several storage types, because the respective storage plugins returned
>>> only the volumes `size` on `volume_size_info` calls, while also the format
>>> is required.
>>>
>>> This patch fixes the issue by returning also `format` and `used`, the
>>> latter also being a non-optional return value for the api call.
>>>
>>> The issue was reported in the forum:
>>> https://forum.proxmox.com/threads/pvesh-get-nodes-node-storage-storage-content-volume-returns-error.123747/
>>>
>>> Signed-off-by: Christian Ebner <c.ebner at proxmox.com>
>>> ---
>>>  PVE/API2/Storage/Content.pm      | 2 +-
>>>  PVE/Storage/ISCSIDirectPlugin.pm | 2 +-
>>>  PVE/Storage/RBDPlugin.pm         | 2 +-
>>>  PVE/Storage/ZFSPoolPlugin.pm     | 2 +-
>>>  4 files changed, 4 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/PVE/API2/Storage/Content.pm b/PVE/API2/Storage/Content.pm
>>> index fe0ad4a..36433ba 100644
>>> --- a/PVE/API2/Storage/Content.pm
>>> +++ b/PVE/API2/Storage/Content.pm
>>> @@ -324,7 +324,7 @@ __PACKAGE__->register_method ({
>>>  
>>>  	my $path = PVE::Storage::path($cfg, $volid);
>>>  	my ($size, $format, $used, $parent) =  PVE::Storage::volume_size_info($cfg, $volid);
>>> -	die "volume_size_info on '$volid' failed\n" if !($format && $size);
>>> +	die "volume_size_info on '$volid' failed\n" if !($format && $size && $used);
>>
>> This change should be it's own patch and I think it's wrong. Can't $used
>> be zero and valid, e.g. for a newly created empty image? You'd either
>> need to use defined($used) or leave it as-is, i.e. not failing and
>> defaulting to 0 when $used is undef. Even with the defined() check, this
>> would break an external plugin that doesn't return $used. While I guess
>> that could be tolerated, we might do it together with the next APIAGE
>> reset and document that volume_size_info explicitly requires it.
> 
> Yes, this was a really sloppy oversight, only added after my testing. On the other hand, should the API call fail for volumes reporting $size = 0 here? Is this intentional or should this also be fixed?

I don't think there are valid cases where $size is 0, or?

>>
>>>  
>>>  	my $entry = {
>>>  	    path => $path,
>>> diff --git a/PVE/Storage/ISCSIDirectPlugin.pm b/PVE/Storage/ISCSIDirectPlugin.pm
>>> index 9777969..eb329d4 100644
>>> --- a/PVE/Storage/ISCSIDirectPlugin.pm
>>> +++ b/PVE/Storage/ISCSIDirectPlugin.pm
>>> @@ -208,7 +208,7 @@ sub volume_size_info {
>>>      my $vollist = iscsi_ls($scfg,$storeid);
>>>      my $info = $vollist->{$storeid}->{$volname};
>>>  
>>> -    return $info->{size};
>>> +    return wantarray ? ($info->{size}, 'raw', 0, undef) : $info->{size};
>>
>> Why return 0 for $used? Doesn't the API call then fail because of the
>> check you added above?
>>
>>>  }
>>>  
>>>  sub volume_resize {
>>> diff --git a/PVE/Storage/RBDPlugin.pm b/PVE/Storage/RBDPlugin.pm
>>> index 9047504..e69c44c 100644
>>> --- a/PVE/Storage/RBDPlugin.pm
>>> +++ b/PVE/Storage/RBDPlugin.pm
>>> @@ -730,7 +730,7 @@ sub volume_size_info {
>>>  
>>>      my ($vtype, $name, $vmid) = $class->parse_volname($volname);
>>>      my ($size, undef) = rbd_volume_info($scfg, $storeid, $name);
>>> -    return $size;
>>> +    return wantarray ? ($size, 'raw', 0, undef) : $size;
>>
>> Also, always returning undef for the parent is also not correct for RBD
>> and ZFS.
> 
> What's the cheapest way here to get the parent for ZFS and RBD?

I think parse_volname() should already give it to you. It's encoded in
the volume ID.