Hi,
so as people asked me on irc, it seems there was a lack of communication
regarding the hardware for the FOSP project
(
https://pagure.io/atomic-wg/issue/153).
Due to internal changes (on top of regular changes) on the other side of
the RH firewall, the acquisition of the servers and the allocation of
the budget took a much longer time than planned, and we got the fund
allocated last week.
So we started working right away on a quote for hardware, and for
various reason, we settled on the following to be shared between the
requirement for my team (OSAS) and Fedora Cloud:
- 1 6U super micro chassis,
https://www.supermicro.com/products/MicroBlade/index.cfm IT can hold 28
blades servers.
So to fill the chassis, we selected:
- 2 * MBI-6418A-T5H, with 4 nodes, each having 4G of ram and a small
disk ( 128 or 64G ssd, depending on price)
- 2 * MBI-6118D-T4, with 4 disk of 1T (or 2T) per disk
- 16 * MBI-6128R-T2, each with 128G of ram, with ssd and 256G if
possible.
and free slots for later (8 slots).
The 4 first blades are to be used by OSAS, to host various small
services and backups. The rest is for Fedora, and the 16 MBI-6128R-T2
would have at least 128G of ram, with 256G ssd, 2 xeon with 18 cores, 1G
connectivity. I did try to optimize to not have too much unused
ressources, but that's gonna still be a lot of ressources.
We plan to host that in the new space we are gonna have near Raleigh, in
the community cage. I do not have a public page to give about that yet,
since we are still working on hosting public pages and a website about
that, but people can ping me if they want more information on this.
For the deployment itself, I had emergency (server, laptop) that did
prevent me from working more on it for now.