HoloMetal

[newgoliath]

Use Judd's datacenter to enable deployment of the holochain docker (and non-docker) work.

[newgoliath]

• [x] get BMC access configured
• [x] add more serial terminal cables (onsite)
• [ ] deploy metal manager (digital rebar) and OS deployer (digital rebar)
• [ ] deploy container orchestrator (openshift/docker-datacenter)
• [ ] deploy metrics and tracing services

[christopherreay]

is there some channel/further discussion of this stuff anywhere

[newgoliath]

@christopherreay - I'm on slack, sometimes. Reach me here, or email. judd@newgoliath.com I need to spend some time reviewing the Dockerfiles build system. Then, once the infra is up and running, we can do continuous builds and integration testing with Jenkins or similar.

[christopherreay]

ok. Eric mentioned your data centre. what goal to you have in mind I general? or specifically? what's the difference between the github Jenkins and what yous are on about?

[newgoliath]

Github offers build/test/deploy/integration test?

[christopherreay]

I sincerely doubt it. are you suggesting something specific? I'm literally asking you what you're suggestion, and you've answered me with a question.

[newgoliath]

I've got some servers at a datacenter that I'm in the midsts of prepping to run a container manager like Kubernetes or Docker Datacenter. I'm interested in deploying the holochains/Docker work there. I'm happy to give access, once I have the gear in shape.

4 x R410 16 core 64gb RAM 7TB HDD

Total: 64 cores 265 GB RAM

servers 1,3,4 model name : Intel(R) Xeon(R) CPU E5640 @ 2.67GHz stepping : 2 microcode : 0x1a cpu MHz : 2660.098 cache size : 12288 KB

server 2 cpu family : 6 model : 26 model name : Intel(R) Xeon(R) CPU E5520 @ 2.27GHz stepping : 5 microcode : 0x19 cpu MHz : 2261.042 cache size : 8192 KB

[zippy]

What's possible on github re CI is Travis, which we are using to test the go code. It will also run single-node app tests (we do it already), but I don't see how we could use that to fire up all the docker instances etc for multi-node testing, which I think is what would be possible as a starter on HoloMetal, not mention holochain app hosting. To me it seems like one thing which will emerge is data centers where people can host subset/copies of the nodes as part of redunancy/resliency for their app ecosystem, and this is port of the ops development we need to explore, to see how this will actually work.

[christopherreay]

OK. Im definately up for some kind of integration testing for everything. Can we make our own badge for the github repo? I think I might put in a bit of work into the GO code testing harness to make it a bit easier to get the results... maybe. Does this mean we are lloking into something that hooks up to github webhooks? Does someone already know how to do that?

OK. I mean Im still up for people having their DHT node on the cloud anyway, and connecting to that from their various devices.

And a thing for me is the app store.... It might make sense for people to be able to spin up small networks of holochain repo's on the store itself (for up to 5 people) or something, to test out if the structure works for them? Or something like that. Perhaps for free up to five people etc... And there may be generic "holochain types" where people can add their specifics (parameterised DNA generation). Not sure how fard that all goes into ceptr rather than holochain.

[christopherreay]

So jenkins itself has a github plugin. Good