Install the new bare metal server(s) for Graph

Orders: https://mybox.inria.fr/smart-link/25b4d73d-381d-4a1c-9d44-7ad4f5ed8161/

Inventory: https://inventory.internal.admin.swh.network/dcim/devices/205/

Summary:

• Management address (DNS): 128.93.134.41 (swh14-adm.inria.fr)
• VLAN configuration: VLAN440
• Management Port: to be defined
• Access Ports: to be defined
• Internal IP(s): 192.168.100.52
• Internal DNS name(s): maxxi.internal.softwareheritage.org

Tasks:

• Declare the servers in the inventory
• Add the management info in the credential store
• Install the OS
• Add puppet configuration (if needed)
• Configure firewall rules (if needed)

changed milestone to %Graph export and graph compression in production [Roadmap - Tooling and infrastructure]

added activity::Deployment hardware labels

marked this issue as related to #4735 (closed)

mentioned in issue #4755 (closed)

marked the checklist item Declare the servers in the inventory as completed

marked the checklist item Add the management info in the credential store as completed

marked the checklist item Install the OS as completed

marked the checklist item Add puppet configuration (if needed) as completed

marked the checklist item Add puppet configuration (if needed) as incomplete

The server is installed, puppet has run, the root password is in the credentials store.

changed the description

JFYI: starting April 3rd we're going to welcome a research engineer who is going to work on the webgraph-rs (Rust) re-implementation. It would be useful to have this machine ready to use by then.

changed the description

It's still not clear to me what exactly is expected to run on this machine, compared to granet or the telecom machine.

The base setup of the machine is done and it's accessible to anyone with an account in the regular SWH puppet manifests. The rest of the setup (for instance, disk layout) is blocked on figuring out what are the requirements for what is expected to run on this machine are.

Notes from the discussions we had with @vlorentz, the initial target was to take control (sys-admin) of the the swh-graph pipeline deployment in order to have a "production ready" instance.

The definitive target is not completely clear as we didn't start the work but all the options are still on the table (bare deployment, elastic deployment, ...).

Opening the accesses to run a poc in this environment was definitely not in the scope. The telecom infra was more the target for that.

To be clear: nobody has mentioned/requested using this new machine (maxxi) to "run a poc".

The last version of swh-graph has been produced (= compressed) entirely on the telecom machine and that is not great in terms of independence of operations for something that is used in production by SWH. So, as a first order of business, the new machine should be configured in a way that allows @vlorentz to fully export/compress/deploy swh-graph using SWH infrastructure only.

Now, the initial plan (dating back maybe to one year ago now) was to use maxxi to compress the graph, and granet to serve it in production. I don't know if it's still realistic to rely on granet for serving the graph, that will need to be verified.

In any event, in terms of disk layout for the compression part (which will for sure need to happen on maxxi), I'm assuming we can just replicate the telecom disk layout as it appears to work well and the machines are identical. But that's up to you folks.

mentioned in commit swh/infra/puppet/puppet-swh-site@1fd317f3

marked the checklist item Add puppet configuration (if needed) as completed

changed the description

closing as the server is installed and managed by puppet.

The graph deployment will be handled in a dedicated issue

closed