[pbs-devel] [PATCH backup/proxmox-backup 0/4] fix #5463: add optional consent banner before login
Gabriel Goller
g.goller at proxmox.com
Thu May 23 14:10:08 CEST 2024
On 23.05.2024 11:24, Thomas Lamprecht wrote:
>Am 23/05/2024 um 09:51 schrieb Dominik Csapak:
>> On 5/22/24 17:31, Thomas Lamprecht wrote:
>>> This is currently still missing any actual barrier as it's all frontend,
>>> shouldn't there be a cookie that is checked on the backend side if a
>>> consent.txt exist? If this specific consent type (RMF AC-8 for US gov)
>>> doesn't need that, it might be worth to replace the generic text box
>>> with a type selection for that, we could always add a "custom" type
>>> that takes a generic text and extent that with an option about how
>>> strict it should be checked, if we get this now.
>>
>> when checking https://csf.tools/reference/nist-sp-800-53/r5/ac/ac-8/
>> (not the "official" document, but very close , the original can be downloaded
>> in docx form here:
>> https://csrc.nist.rip/projects/risk-management/about-rmf/assess-step/assessment-cases-download-page)
>> it does not seem to be necessary for any cookie handling
>> since it just wants the disclaimer to be displayed before login
>
>ack, thanks for the links.
>
>>
>>>
>>> And how should API calls made using API tokens get handled, should they
>>> have a header signalling consent or not? If, should there be a set of
>>> standard consents that one can explicitly consent too? As a blanket
>>> consent to an unknown text would not be of much use.
>>
>>
>> also it says that this is only for human interaction, so any api
>> access etc. is exempt IIUC
>
>
>So pretty much a worthless "keep out" sign [0], can one be even
>enterprise ready without those? ;-)
>
>[0]: https://i.imgur.com/mSHi8.jpeg
>
>Anyhow, fine by me, but I then still would prefer having this saved
>as structured data with an explicit type so that we can easily extend
>this with an option for actually enforcing such a consent, if ever
>requested.
>
>Maybe we can even add it as encoded text to an existing config, for PVE
>the datacenter one would be a good fit, for PMG with also have a cluster
>wide one IIRC and for PBS we could just add it to the node.cfg (and cache
>inside the http daemon).
I don't think we will gain much from adding the text in a config file
here. The config files don't support multi-line values and thus we have to
escape all the newlines. If we do this, we would have to introduce a ui
textfield where the user can edit the consent file, otherwise he would
have to escape all the newlines manually in the node.cfg file (which is
a PITA).
I am also kind of opposed to a ui element because this is quite a niche
feature and would only clog the interface.
I don't think an option to strictly enforcing the consent won't come any
time soon, as it's quite complicated to implement and is mostly used as
a legal requirement anyway.
Another plus would be that VMWare does the same, so a user would just
have to come the .txt file /etc/proxmox-backup (+ rename it) and would
be ready to go.
If we want to add other visual configs such as optional buttons (e.g.:
"I Accept", "I Decline") I think we should add the in the txt file like
such (or something similar):
multi-line
text
I Agree|I Disagree|Ignore
More information about the pbs-devel
mailing list