RAS API Workstream: Difference between revisions

From OpenCompute
Jump to navigation Jump to search
No edit summary
No edit summary
 
(3 intermediate revisions by 2 users not shown)
Line 15: Line 15:


Having no industry standard for RAS management makes each HW vendor solution specific and resource intensive; while delaying RAS feature adoption. <br />
Having no industry standard for RAS management makes each HW vendor solution specific and resource intensive; while delaying RAS feature adoption. <br />
When creating a specification to solve the standardization problem some major pain points need to be considered:
When creating a specification to solve the standardization problem some major pain points need to be considered:
* Specification for  common '''abstraction''' for RAS  between software and hardware that is '''vendor agnostic''' as well as '''platform agnostic'''.
* Specification for  common '''abstraction''' for RAS  between software and hardware that is '''vendor agnostic''' as well as '''platform agnostic'''.
* Connectivity for RAS management should allow '''co-existence''' of In Band and Out of Band methods
* Connectivity for RAS management should allow '''co-existence''' of In Band and Out of Band methods
Line 24: Line 26:


The current baseline is [https://docs.google.com/document/d/1LhGtwIdhw7XuH84rPfRRq8grhl-fsIN1/edit?usp=share_link&ouid=115560947677973489092&rtpof=true&sd=true Revision 0.8].  We plan to meet with companies that agree with this objective in order to draft a complete specification that solves the problem statement.  
The current baseline is [https://docs.google.com/document/d/1LhGtwIdhw7XuH84rPfRRq8grhl-fsIN1/edit?usp=share_link&ouid=115560947677973489092&rtpof=true&sd=true Revision 0.8].  We plan to meet with companies that agree with this objective in order to draft a complete specification that solves the problem statement.  


==Workstream Leadership==
==Workstream Leadership==


 
* [mailto:acwalton@google.com Drew Walton] (Google)
:- [mailto:acwalton@google.com Drew Walton] (Google)[mailto:antonio.j.hasbun.marin@intel.com Antonio Hasbun] (Intel)
* [mailto:antonio.j.hasbun.marin@intel.com Antonio Hasbun] (Intel)


==Get Involved==
==Get Involved==


The process to join the workstream is as follows:
The process to join the workstream is as follows:
<br />
 
# Join the mailing list: [mailto:OCP-RAS-API+subscribe@OCP-All.groups.io OCP-RAS-API+subscribe@OCP-All.groups.io]
* Join the mailing list: [mailto:OCP-RAS-API+subscribe@OCP-All.groups.io OCP-RAS-API+subscribe@OCP-All.groups.io]
# Join the [https://drive.google.com/drive/folders/0AKz7Api2DyIRUk9PVA Google drive] This will grant you access to the Google drive for the workstream and we can download and get the CLA signed
* Join the [https://drive.google.com/drive/folders/0AKz7Api2DyIRUk9PVA Google drive] This will grant you access to the Google drive for the workstream and we can download and get the CLA signed
<br/>
 
This is a link to the CLA that needs to be signed in order to participate in the workstream. [https://docs.google.com/document/d/1LhGtwIdhw7XuH84rPfRRq8grhl-fsIN1/edit?usp=share_link&ouid=115560947677973489092&rtpof=true&sd=true RAS API CLA  ]
This is a link to the CLA that needs to be signed in order to participate in the workstream. [https://docs.google.com/document/d/1LhGtwIdhw7XuH84rPfRRq8grhl-fsIN1/edit?usp=share_link&ouid=115560947677973489092&rtpof=true&sd=true RAS API CLA  ]
<br />


===Workstream Meetings require a signed CLA, please add yourself to the communication channels to get the invitation after signing the CLA. ===


===Workstream Meets every TBD  ===
===Mailing List ===
: - [https://global.gotomeeting.com/join/  Link to the Meeting]
:  - You can also dial in using your phone : US (Toll Free): 1 877 309 2073 / US: +1 (646) 749-3129 - (Access Code: )


===Mailing List ===
Participate in the discussion, mailing list:  OCP-RAS-API@OCP-All.groups.io [https://ocp-all.groups.io/g/OCP-RAS-API Mailing List Info]
Participate in the discussion, mailing list:  OCP-RAS-API@OCP-All.groups.io [https://ocp-all.groups.io/g/OCP-RAS-API Mailing List Info]


==Links to Documents==
==Links to Documents==


===[https://drive.google.com/drive/folders/0AKz7Api2DyIRUk9PVA Work in Progress Documents Folder]:===
* [https://drive.google.com/drive/folders/0AKz7Api2DyIRUk9PVA Work in Progress Documents Folder]:===
 
 


===Presentations & Other Documents===
===Presentations & Other Documents===
OCP Presentation Template - please contact [mailto:michael@opencompute.org Michael Schill] or [mailto:archna@opencompute.org Archna Haylock] for a copy


==Past Meeting Recordings==
OCP Presentation Template - please contact [mailto:michael@opencompute.org Michael Schill] for a copy

Latest revision as of 22:37, 1 May 2024

OCP-Open-compute-hardware-fault-management-icon-3x-v1-1a.png

Welcome[edit]

Welcome to the OCP RAS API workstream. This is a CLA workstream of the OCP Hardware Management Project
This workstream is under CLA so in order to participate we need to sign a RAS API CLA .
Disclaimer: Please do not submit any confidential information to the Project Community.
All presentation materials, proposals, meeting minutes and/or supporting documents can be utilize for the upcoming RAS API specification.
If you have any questions please contact OCP.

Workstream Overview[edit]

This workstream will be working on creating the RAS API specification.

Problem Statement[edit]

Having no industry standard for RAS management makes each HW vendor solution specific and resource intensive; while delaying RAS feature adoption.

When creating a specification to solve the standardization problem some major pain points need to be considered:

  • Specification for common abstraction for RAS between software and hardware that is vendor agnostic as well as platform agnostic.
  • Connectivity for RAS management should allow co-existence of In Band and Out of Band methods
  • Logging mechanism must solve RAS unique problems (Error storms, error sampling, etc.)
  • RAS feature abstraction should be generic enough to have open source drivers that can do a reasonable job at managing RAS;
    while leaving the opportunity for hardware vendor to differentiate and keep innovating.

Methodology[edit]

The current baseline is Revision 0.8. We plan to meet with companies that agree with this objective in order to draft a complete specification that solves the problem statement.

Workstream Leadership[edit]

Get Involved[edit]

The process to join the workstream is as follows:

This is a link to the CLA that needs to be signed in order to participate in the workstream. RAS API CLA

Workstream Meetings require a signed CLA, please add yourself to the communication channels to get the invitation after signing the CLA.[edit]

Mailing List[edit]

Participate in the discussion, mailing list: OCP-RAS-API@OCP-All.groups.io Mailing List Info

Links to Documents[edit]

Presentations & Other Documents[edit]

OCP Presentation Template - please contact Michael Schill for a copy