This task will outline the overall steps needed (and link sub-tasks) for actionable items from the operations teams breakout group on Hardare provisioning & automation.

The notes of the session are listed on https://office.wikimedia.org/wiki/Operations/Operations_Meeting_Notes/2015-10-12_Ops_Offsite/hardware-automation-workflow

Action Items:

• servermon fully in production (big one)
  • related T88424: Migrate racktables to servermon
  • related T84001: alternatives to racktables ?
  • mac address and network port info missing (among others)
• easier allocation of IPs for management (in dns.git)
  • this is easier for codfw, since its properly ordered. eqiad is ordered by service group and horrible to work with.
    • we should clean up and renumber eqiad when this is automated.
  • the same concept can be applied to add/remove machines from puppet.git (e.g. dhcp entries)
• PXE-boot linux image to run administration tasks
  • related T78135: Provide a pxe-bootable rescue image
  • ran only from a dedicated "install vlan"
• investigate ssh keys for idrac/ilo
  • related {T113557}
• lock mgmt vlan from non ops bastions
  • {T79294} OLD rtimport task for this.

Additionally, there has long been discussion about scripting all mgmt tasks.

• Interesting link for HP mgmt wrapper in python: https://github.com/seveas/python-hpilo