1 |
On Sun, 26 Apr 2020 10:08:32 +0200 |
2 |
Michał Górny <mgorny@g.o> wrote: |
3 |
|
4 |
> A proper solution to cluster problem would probably involve some way to |
5 |
> internally collect and combine data data before submission. If you have |
6 |
> large clusters of similar systems, I think you'd want to have all |
7 |
> packages used on different systems reported as one entry. |
8 |
|
9 |
For this, I'd suggest the ability to have an overrideable |
10 |
"STATS_SERVER" (or something) ENV var URI that tells the submission |
11 |
clients where to send their reports to. |
12 |
|
13 |
Then have some server shipped in gentoo people can deploy, and submit |
14 |
aggregated as a cron job, or potentially hand review the aggregated |
15 |
submission data before submission, and potentially have tools to |
16 |
whittle data out you don't want to share at the org level. |
17 |
|
18 |
Such a tool is potentially useful to an organisation even without its |
19 |
"submit to gentoo" capacity, as being able to internally analyse what |
20 |
your organisation is using seems to be useful. |
21 |
|
22 |
(eg: provide an admin a single point of information showing what |
23 |
packages they need to audit, if all the nodes in the org are not |
24 |
entirely controlled at the top level) |
25 |
|
26 |
Though I think the overall design of anonymity by design is useful, I |
27 |
can see usecases, especially in the organisation model, where being |
28 |
able to voluntarily self-identify a node could be useful without |
29 |
inherently being a privacy concern. |
30 |
|
31 |
And you'd configure your relay to suppress these node identities in the |
32 |
submitted data, or map them to a different org-wide identity. |
33 |
|
34 |
Example: |
35 |
I need to find somebody who is using <x> so I can ask them if <y> |
36 |
works, or if <z> is important to this package. |
37 |
|
38 |
Example: |
39 |
Data indicates somebody within my org is using <x>, and I need to ask |
40 |
them not to use <x>, as its licensing terms are not compatible with |
41 |
our org. |
42 |
|
43 |
Though for cases of voluntary identification, you'd need an interface |
44 |
on the server node somewhere that allows you to generate unique ident |
45 |
tokens, and associate data with them, possibly with a list of flags |
46 |
dictating what records associated with this identity may be used for |
47 |
(eg: Contact [y/n] ) |