OMPI_CRCP(7)

πŸ“¦
version 4.1.6 Β· 5 versions
, in alpine 3.24, ubuntu noble

πŸ’– OMPI_CRCP πŸ’–

okay bestie, welcome to the OMPI_CRCP manual page for Open MPI β€” this is the Open MPI MCA Checkpoint/Restart Coordination Protocol (CRCP) Framework</p> ✨ It's the overview of Open MPI's CRCP framework plus the selected modules, and it's sitting pretty here at Open MPI 4.1.6. Let's get into it, no cap 🫢


🌈 DESCRIPTION 🌈

The CRCP Framework is what Open MPI uses to encapsulate all the different Checkpoint/Restart Coordination Protocols out there β€” think Coordinated, Uncoordinated, Message/Communication Induced, ... you name it πŸ’… It's giving "one framework to hold them all" energy.


πŸŽ€ GENERAL PROCESS REQUIREMENTS πŸŽ€

So if a process wants to use the Open MPI CRCP components, it has to follow a few programmatic requirements, fr fr. Like, the vibes have to be right:

  • First off, the program must call MPI_INIT early in its execution. Early. Not later. ⏰
  • The program must call MPI_FINALIZE before termination. Do not skip this, bestie, it's not optional 😭
  • A user can initiate a checkpoint of a parallel application by using the ompi-checkpoint(1) and ompi-restart(1) commands πŸ”₯

✨ AVAILABLE COMPONENTS ✨

Open MPI currently ships with one CRCP component: coord. That's it. That's the whole roster. πŸ‘‘

The following MCA parameters apply to all components:

  • crcp_base_verbose β€” sets the verbosity level for all components. Default is 0, which means silent except on error. So if you want the tea, you gotta turn it up πŸ“

🌸 coord CRCP Component 🌸

The coord component implements a Coordinated Checkpoint/Restart Coordination Protocol similar to the one implemented in LAM/MPI. Iconic behaviour, honestly.

The coord component has the following MCA parameters:

  • crcp_coord_priority β€” the component's priority to use when selecting the most appropriate component for a run πŸ‘‘
  • crcp_coord_verbose β€” sets the verbosity level for this component. Default is 0, or silent except on error 🀫

πŸ¦‹ none CRCP Component πŸ¦‹

The none component simply selects no CRCP component. All of the CRCP function calls return immediately with OMPI_SUCCESS. Chillest component in the building, honestly πŸ’€

This component is the last component to be selected by default. What that means: if another component is available, and the none component was not explicitly requested, then Open MPI will attempt to activate all of the available components before falling back to this component. So it's the backup plan, the safety net, the "if all else fails" bestie 🫢

lowkey important: none only shows up if you didn't explicitly ask for it AND everything else got a shot first. iykyk.

πŸ‘‘ SEE ALSO πŸ‘‘

ompi-checkpoint(1), ompi-restart(1), opal-checkpoint(1), opal-restart(1), orte_snapc(7), orte_filem(7), opal_crs(7)


and that's the whole page, bestie!! πŸŽ€ this was Open MPI 4.1.6, dated Sep 30, 2023, and honestly? she ate. go forth and checkpoint responsibly πŸ’…βœ¨πŸŒˆ

πŸ“¦ where this page comes from