Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Infrastructure for Orka (2024 and beyond) #3686

Open
9 of 12 tasks
UlisesGascon opened this issue Apr 19, 2024 · 17 comments
Open
9 of 12 tasks

Infrastructure for Orka (2024 and beyond) #3686

UlisesGascon opened this issue Apr 19, 2024 · 17 comments

Comments

@UlisesGascon
Copy link
Member

UlisesGascon commented Apr 19, 2024

I plan to work on it during the weekend, so I can provide a good overview on the next build meeting on Tuesday.

Current tasks on MacOS infra

Blocked until ARM nodes are provided

  • Confirm org decision regarding new ARM nodes (discussion ongoing in the mailing list)
  • Add new VMs for MacOS 13 ARM
  • Add new VMs for MacOS 11 ARM
@UlisesGascon
Copy link
Member Author

Current Orka state

updated on April 19, 2024

SSH port Node: macpro-4 Node: macpro-5 Node: macpro-6
8822 release-macos11-x64-1 empty test-macos11-x64-1
8823 empty empty test-macos11-x64-2
8824 empty test-macos1015-x64-2 test-macos1015-x64-1
8825 empty empty empty

@UlisesGascon
Copy link
Member Author

UlisesGascon commented Apr 19, 2024

Next Orka state

updated on April 22, 2024

Intel Nodes

SSH port Node: macpro-4 Node: macpro-5 Node: macpro-6
8822 release-macos11-x64-1 test-macos13-x64-2 test-macos11-x64-1
8823 test-macos13-x64-1 release-macos13-x64-1 test-macos11-x64-2
8824 empty test-macos1015-x64-2 test-macos1015-x64-1
8825 empty empty empty

ARM Nodes

We assume that ARM Nodes can handle only 2 VMs and not +4 as Intel in the past due license limitations. This needs to be confirmed with support AFAIK?

SSH port Node: arm-1 Node: arm-2 Node: arm-3
8822 test-macos11-arm64-1 release-macos13-arm64-1 empty
8823 release-macos11-arm64-1 test-macos13-arm64-1 test-macos13-arm64-2

How Nearform machines are "relocated"?

  • release-nearform-macos11.0-arm64-1 -> release-orka-macos11-arm64-1
  • test-nearform-macos11.0-arm64-1 -> test-orka-macos11-arm64-1

@targos
Copy link
Member

targos commented Apr 22, 2024

release-macos13-x64-2
release-macos13-arm64-2

I don't think it's necessary to have two identical release machines.

@targos
Copy link
Member

targos commented Apr 22, 2024

test-nearform-macos11.0-arm64-1

Are these typos?

@UlisesGascon
Copy link
Member Author

UlisesGascon commented Apr 22, 2024

Great feedback @targos! I updated the tables

I don't think it's necessary to have two identical release machines.

We have space for redundancy, but let's remove them for now.

Are these typos?

I made a better reference for the "relocated" machines

@targos targos pinned this issue May 2, 2024
@targos
Copy link
Member

targos commented May 2, 2024

release-macos13-x64-2
release-macos13-arm64-2

I don't think it's necessary to have two identical release machines.

Actually, I think we should have one x64 and two arm64 machines, because there are two jobs that run on macos-arm64 during a release (osx11-release-pkg and osx11-arm64-release-tar).

@ryanaslett
Copy link
Contributor

Some questions/thoughts/suggestions:

  1. Requirements Question: Do we still need to support 10.15 and/or 11? from (https://github.com/nodejs/node/blob/main/BUILDING.md#supported-platforms) I see:

Node.js does not support a platform version if a vendor has expired support for it. In other words, Node.js does not support running on End-of-Life (EoL) platforms. This is true regardless of entries in the table below.

And the table lists MacOS 11>.

And that table may be outdated as it seems as though MacOS 11 was EOL as of November 2023 ?

  1. ARM support in Orka:

We assume that ARM Nodes can handle only 2 VMs and not +4 as Intel in the past due license limitations. This needs to be confirmed with support AFAIK?

https://orkadocs.macstadium.com/docs/apple-arm-based-support confirms this:

IMPORTANT

You can deploy up to 2 VMs per Apple silicon-based node.

  1. From what I can gather macOS infra seems to be brittle, with nodes often running into disk issues/maintenance issues.

#3592
#3685
(https://github.com/nodejs/build/issues?q=is%3Aissue+macos+is%3Aclosed+disk) etc.

My suggestion to avoid Jenkins worker decay is to lean into an ephemeral node strategy so that each build has a fresh Orka instance to run on.

We can do that with the following Jenkins plugin for Orka:
https://plugins.jenkins.io/macstadium-orka/#plugin-content-ephemeral-agents

We would first need to set up a packer build process to create our VM images so that Orka would have a baseline image to create:
https://orkadocs.macstadium.com/docs/packer

The packer process can leverage our existing ansible playbooks:
https://developer.hashicorp.com/packer/integrations/hashicorp/ansible/latest/components/provisioner/ansible.

This strategy would require that we have an Orka3.0 cluster. Rather than trying to do an upgrade of the existing cluster, I propose that we ask macstadium to allow us to provision a new cluster with the resources we need in it (enough arm/intel backing nodes for our macos11/13 testing and release), get it built/provisioned and working, and then decommission/return all the existing macstadium/orka machines.

I believe this would end up with us using roughly the same amount of resources, so should be palatable for macstadium to support this transition.

@mhdawson
Copy link
Member

This strategy would require that we have an Orka3.0 cluster. Rather than trying to do an upgrade of the existing cluster, I propose that we ask macstadium to allow us to provision a new cluster with the resources we need in it (enough arm/intel backing nodes for our macos11/13 testing and release), get it built/provisioned and working, and then decommission/return all the existing macstadium/orka machines.

+1 from me if Macstadium will support that

@UlisesGascon
Copy link
Member Author

UlisesGascon commented Aug 9, 2024

Quick update from our last call with MacStadium:

Next week we will have a new Orka cluster (v3) that includes 2 nodes (Intel and ARM):

  • Mac Studio - G1MC M1M/10/32/16/64GB/2TB/10G
  • Mac mini G4E - i7/3.2Ghz/6C/64G/1T/SSD/10G

Pending:

Dependencies

✅ Setup Jenkins <-> Orka

Current status: Completed.

✅ Create Image templates

Current status: Completed.

✅ Trigger Ephemeral VMs from Jenkins

  • Setup Ephemeral nodes from Jenkins on demand for the test CI and probe that they can build and test node.
  • Setup Ephemeral nodes from Jenkins on demand for the release CI and probe that they can build, test and sign node.

Current status: Completed

Jobs and Agents Migration

  • Add the MacOS13 to the ci test existing jobs (commit-test, night builds, v8, CITGM...) to check that all the current setup can run all the pipelines
  • Add the MacOS13 to the ci release existing jobs to check that all the current setup can run all the pipelines
  • Configure Jenkins Plugin to be compatible with MacOS11 and MacOS10.15 -mmacosx-version-min (see: Proposal: Drop support for MacOS prior to 13 #3876) in the test ci
  • Configure Jenkins Plugin to be compatible with MacOS11 and MacOS10.15 -mmacosx-version-min (see: Proposal: Drop support for MacOS prior to 13 #3876) in the release ci
  • Remove labels from legacy jenkins agents and check that the jobs are working as expected in the test ci
  • Remove labels from legacy jenkins agents and check that the jobs are working as expected in the release ci

Current status: @UlisesGascon working on the setup.

Clean up

Other

Deadline
The idea is to try to achieve this transition in 30 days.

Important

We don't expect any downtime will doing the migration as we will have a new cluster working on isolation will the current system is in place until we are ready to transfer the operations to the new cluster and then decommission the HW.

Challenges

  • Lack of support for MacOS 11 ARM in Orka: details
  • Cannot deploy more than 2 VMs on an ARM host.
  • Error: admission webhook "vimage.kb.io" denied the request: cannot delete image "macos13-intel-test-latest.img". The image is being used by one or more VMs: vm-ttdzh. Remove the VMs and try again
  • When Jenkins create a cloud agent if this agent failed then is not removed from the cluster
  • Evaluate how to use namespace
  • New VMs are much slower than expected (@UlisesGascon investigating)
  • Unify the HCL templates into a single one if possible (see: Orka template updates #3906 (comment))

UlisesGascon added a commit that referenced this issue Aug 31, 2024
UlisesGascon added a commit that referenced this issue Aug 31, 2024
- Related: #3686
- Updated documentation with the new environmental variables
- Added Packer template for Orka MacOS 13 ARM test
- Updated CI to use all the expected environmental variables
UlisesGascon added a commit to UlisesGascon/build that referenced this issue Aug 31, 2024
UlisesGascon added a commit to UlisesGascon/build that referenced this issue Aug 31, 2024
UlisesGascon added a commit to UlisesGascon/build that referenced this issue Aug 31, 2024
UlisesGascon added a commit that referenced this issue Sep 2, 2024
@UlisesGascon
Copy link
Member Author

UlisesGascon commented Sep 16, 2024

Current status

I will be on PTO from the 19th to the 25th. I made some changes to the templates to add the missing dependencies (#3906).

So, @ryanaslett, in case you want to help with this during my time off:

  • Check why the iojs+release-Ulises-test-orca is not passing. The current error (10:09:52 Makefile:1030: *** No xz command, cannot continue. Stop.) (details) is related to the PATH (I think), as xz is included on all the machines already since my last PR.
  • Try to modify the base image for release and test machines to include and use Xcode 15.2, based on the discussion with @targos. Documentation
  • Keep running and patching the other pipelines that we cloned with @mhdawson.

Probably the next errors in the CI will be related to the users; currently, we only have the admin user. Maybe we need to create a separate one like iojs in order to make the CI pipelines work.

@ryanaslett
Copy link
Contributor

Check why the iojs+release-Ulises-test-orca is not passing. The current error (10:09:52 Makefile:1030: *** No xz command, cannot continue. Stop.) (details) is related to the PATH (I think), as xz is included on all the machines already since my last PR.

Started in on this.

The PATH variable is set on the existing macos machines via the script that launches the jenkins agent:
This template:
https://github.com/nodejs/build/blob/main/ansible/roles/jenkins-worker/templates/start.j2#L10
Creates a script here:
https://github.com/nodejs/build/blob/main/ansible/roles/jenkins-worker/tasks/main.yml#L179-L185
And this Template:
https://github.com/nodejs/build/blob/main/ansible/roles/jenkins-worker/templates/org.nodejs.osx.jenkins.plist
Gets put into /Library/LaunchDaemons
https://github.com/nodejs/build/blob/main/ansible/roles/jenkins-worker/vars/main.yml#L33-L37

I've added ARCH, DESTCPU, and PATH to the Environment variables to the Orka Cluster Cloud Template configurations on ci-release machine.

The osx13-x64-release-tar job worked and signed the tarball, but failed to push the release to node-www, so, need to adjust that next.

image

@targos
Copy link
Member

targos commented Oct 3, 2024

@richardlau
Copy link
Member

We need this config in the image: https://github.com/nodejs/build/blob/main/ansible/roles/release-builder/files/ssh_config

node-www also has a ufw2 firewall and will not allow connections from ip addresses not on the allowlist.

@ryanaslett
Copy link
Contributor

I've added the main orka address to the ufw2 firewall on node-www (199.7.167.98) I've confirmed that this is the address that ephemeral nodes will all appear as to node-www.

@ryanaslett
Copy link
Contributor

I've requested the new nodes from MacStadium to fill out the rest of our capacity, and got a response today that they are aiming to have the nodes installed by Wed, Oct 30th.

@mhdawson
Copy link
Member

Great to see the details and progress on this front.

One thought is that once everything is landed it would be great to do a deep dive session for other build team members who are interested in learning a bit more about now it works.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

5 participants