1 |
On Thu, Sep 6, 2018 at 11:22 AM Matthew Thode <prometheanfire@g.o> wrote: |
2 |
> |
3 |
> One idea I had infra-wise was to expand the idea of our devspace. |
4 |
> Increase the footprint of our virtualizatoin infra and (on request) give |
5 |
> devs a VM for dev work. Possibly even do some arm64 or ppc64 work there |
6 |
> too. Beyond that, running a binhost may be an idea, get people |
7 |
> bootstrapped faster. |
8 |
|
9 |
If anything I think it would make more sense to try to decentralize |
10 |
things more, to let individual contributors run more Gentoo services |
11 |
off of their own stuff. For example, it would be useful if a service |
12 |
hosted on a dev machine could allow LDAP users to authenticate in a |
13 |
secure manner, and so on. |
14 |
|
15 |
The idea would be to make us less dependent on centrally-owned |
16 |
infrastructure, so that it is less of a disaster if we lose access to |
17 |
servers for whatever reason. It would also reduce our operating |
18 |
costs. Services hosted by devs could still be FOSS with published |
19 |
configurations/sources/etc so that they're easily replicated. |
20 |
|
21 |
The biggest problem is that some of our key infra isn't easily |
22 |
distributed, like bugzilla. LDAP is of course by its nature a |
23 |
centralized service (even with something more federated you need some |
24 |
kind of standard of trust unless we do the web-of-trust thing - I |
25 |
think that needs to be down the road). Until somebody creates a |
26 |
git-like distributed bugzilla solution we'll probably need some kind |
27 |
of central repository, especially if we have private bugs such that we |
28 |
can't just publish the database and let anybody replicate it. |
29 |
|
30 |
-- |
31 |
Rich |