RISC-V system migration at PLCTlab

[sxa]

A number of the existing machines at the PLCTlab have been replaced with systems running Ubuntu 22.04 in a different lab. They will require the playbooks to be re-run on them.

Some notes regarding things that will need to be done on them:

• openjdk-8-jdk in the Common playbook role is not available (unsurprisingly!)
• docker.io should be installed instead of docker-ce in the Docker role (possibly true for all recent Ubuntus)
• The gcc_ roles need to be commented out as we do not currently have a full set of gcc tarballs available for RISC-V and the roles are not set up to skip it.

I've also been skipping:

• the swap_file role when trying it as the 16GB machines seem to run ok without a swapfile in place.
• nagios_master_config because I was running the playbook locally against localhost and provider wasn't defined (perhaps need to be checked in the when clause of that role)
• nagios_tunnel because it needs ansible_port

Noting that I saw most of these when attempting to set up one of my own systems with 23.04.

[sxa]

First system connected and live at https://ci.adoptium.net/computer/build-plctlab-ubuntu2204-riscv64-1/ and now running a build job to verify it. (-0 was connected after I ran a test on the machine but then fell off the network so will need to be restarted). Note that -1 has been allowed to create a 2Gb swap file via the ansible role so swap_file wasn't skipped on that one.

Plan will be to get another 5 which have been allocated to us live and all set up the same way.

[sxa]

$ cat /unmatched302
test-plctlab-ubuntu2204-riscv64-302 ansible_ssh_common_args='-F /root/.ssh/config'
$ ansible-playbook -i /unmatched302 --skip-tags=nagios_master_config,nagios_tunnel playbooks/AdoptOpenJDK_Unix_Playbook/main.yml

Or on another system without passwordless sudo:

$ cat /unmatched303
test-plctlab-ubuntu2204-riscv64-303 ansible_ssh_common_args='-o ProxyCommand="ssh rvlab -W %h:%p"'
$ ansible-playbook -b -u sxa -i /unmatched303 --skip-tags=nagios_master_config,nagios_tunnel playbooks/AdoptOpenJDK_Unix_Playbook/main.yml

Or using a full set of commands without a custom ssh config file:

$ cat /unmatched304
[plctlab]
test-plctlab-ubuntu2204-riscv64-304 port=13304

[plctlab:vars]
ansible_user=sxa
ansible_host=127.0.0.1
ansible_ssh_common_args='-o ProxyCommand="ssh [email protected] -W %h:%p"'
$ ansible-playbook -K -i /unmatched304 --skip-tags=nagios_master_config,nagios_tunnel playbooks/AdoptOpenJDK_Unix_Playbook/main.yml

Or with a yaml file:

$ cat /unmatched305
all:
   hosts:
      plctlab-ubuntu2204-riscv64-305 {port: 13305}
      plctlab-ubuntu2204-riscv64-312 {port: 13312}
   vars:
      ansible_user: sxa
      ansible_host: 127.0.0.1
      ansible_ssh_common_args: -o ProxyCommand="ssh [email protected] -W %h:%p"
$ ansible-playbook -K -i /unmatched305 --skip-tags=nagios_master_config,nagios_tunnel playbooks/AdoptOpenJDK_Unix_Playbook/main.yml

[sxa]

The job at https://ci.adoptium.net/job/SXA_plct_speedtest/ is running regular testing to verify that the speed of the machines is adequate. It was doing well at the beginning of this year but seems to have fallen back to random slowness.