May 2022 - mpiwg-ft - mailman3.cels.anl.gov

Re: [mpiwg-ft] WG FT Meeting schedule
by Aurelien Bouteiller 16 May '22

16 May '22

Hi everyone, We have a meeting scheduled for today. It may be short, as the agenda for next week Forum Meeting is locked. I have announced the following readings for the Forum Meeting next week. * expansion of error handler association to cover sessions (and clarifications) * https://github.com/mpi-forum/mpi-standard/pull/644 * fine-grain recovery (slice 1) (aka ULFM v2) * https://github.com/mpi-forum/mpi-standard/pull/665 We’ll make a roundtable call to see if someone wants to add an agenda item. Aurelien > On May 2, 2022, at 12:44, Aurelien Bouteiller <bouteill(a)icl.utk.edu> wrote: > > Meeting Summary > =============== > > > Advice about unsupported fault types > ------------------------------------ > > We discussed changes to the introduction of fine-grain recovery to add an advice to implementors about non-process-fail-stop faults. Text visible here: > > https://github.com/mpiwg-ft/mpi-standard/pull/17 > > Overall text appears clear enough, please add more feedback on the PR itself (ping me if you do not have access to the repo). > > > All PRs with “error handling” components > ---------------------------------------- > > Here is a list of all the PRs to the main standard that do touch error handling. I don’t see any issues with them, but please bring forward any reservations you may have for the next meeting. > > * Sessions and LAST_USED_CODE: https://github.com/mpi-forum/mpi-standard/pull/657 > * Sessions and language bindings: https://github.com/mpi-forum/mpi-standard/pull/658 > * MPI Objects, Sessions and where to find an error handler: https://github.com/mpi-forum/mpi-standard/pull/644/files > * MPI_Comm_create_from_group errhandler to invoke: https://github.com/mpi-forum/mpi-standard/pull/629 > * Tools clarify they use return codes (not MPI error codes): https://github.com/mpi-forum/mpi-standard/pull/646 > * New error class MPI_ERR_ERRHANDLER: https://github.com/mpi-forum/mpi-standard/pull/633 > * MPI_DELETE_ERROR_CLASS and friends: https://github.com/mpi-forum/mpi-standard/pull/166 > > > Next Meeting > ============ > > Joint with Session: Monday May 9th 13:00 EST > WG only: Monday May 16th, 12:00 EST > > Best Regards, > Aurelien > > -- > Aurelien Bouteiller, Ph.D. > Innovative Computing Laboratory; The University of Tennessee; > 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA > +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) > > > > >> On May 2, 2022, at 11:39, Ignacio Laguna via mpiwg-ft <mpiwg-ft(a)lists.mpi-forum.org> wrote: >> >> I have conflict today and can't attend. I will call in next time. >> Thanks >> Ignacio >> >> On 5/2/22 8:38 AM, Aurelien Bouteiller via mpiwg-ft wrote: >>> As planned we will have a meeting today at noon EST. >>> On the agenda are >>> 1. changes to fine-grain recovery (aka ulfm v2) from past week plenary >>> 2. Quick review of all PR’s touching error handling >>> Best, >>> Aurelien >>>> On Apr 11, 2022, at 10:37, Aurelien Bouteiller <bouteill(a)icl.utk.edu> wrote: >>>> >>>> Hey all, >>>> >>>> We will have a joint meeting today with the Sessions WG (connexion info is on their GitHub) at 1pm EST. >>>> >>>> We will have the solo WG Errors & FT meeting next week at noon EST. We will review all current PRs that touch error handling. >>>> >>>> >>>> In order to clarify our schedule, I have created a new calendar invite with the correct recurrence, you can download it from the WG wiki. https://urldefense.us/v3/__https://github.com/mpiwg-ft/ft-issues/wiki__;!!G… >>>> >>>> TO SUMMARIZE: THE WG ERRORS & FT MEETINGS WILL BE EVERY 2 WEEKS AT NOON EST, STARTING MON APR. 18. >>>> >>>> Best Regards, >>>> Aurelien >>>> >>>> -- >>>> Aurelien Bouteiller, Ph.D. >>>> Innovative Computing Laboratory; The University of Tennessee; >>>> 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA >>>> +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) >>>> >>>> >>>> >>>> >>>> >>>> >>> _______________________________________________ >>> mpiwg-ft mailing list >>> mpiwg-ft(a)lists.mpi-forum.org >>> https://urldefense.us/v3/__https://lists.mpi-forum.org/mailman/listinfo/mpi… >> _______________________________________________ >> mpiwg-ft mailing list >> mpiwg-ft(a)lists.mpi-forum.org >> https://lists.mpi-forum.org/mailman/listinfo/mpiwg-ft > > > > > > >

1 0

Re: [mpiwg-ft] WG FT Meeting schedule
by Aurelien Bouteiller 02 May '22

02 May '22

Meeting Summary =============== Advice about unsupported fault types ------------------------------------ We discussed changes to the introduction of fine-grain recovery to add an advice to implementors about non-process-fail-stop faults. Text visible here: https://github.com/mpiwg-ft/mpi-standard/pull/17 Overall text appears clear enough, please add more feedback on the PR itself (ping me if you do not have access to the repo). All PRs with “error handling” components ---------------------------------------- Here is a list of all the PRs to the main standard that do touch error handling. I don’t see any issues with them, but please bring forward any reservations you may have for the next meeting. * Sessions and LAST_USED_CODE: https://github.com/mpi-forum/mpi-standard/pull/657 * Sessions and language bindings: https://github.com/mpi-forum/mpi-standard/pull/658 * MPI Objects, Sessions and where to find an error handler: https://github.com/mpi-forum/mpi-standard/pull/644/files * MPI_Comm_create_from_group errhandler to invoke: https://github.com/mpi-forum/mpi-standard/pull/629 * Tools clarify they use return codes (not MPI error codes): https://github.com/mpi-forum/mpi-standard/pull/646 * New error class MPI_ERR_ERRHANDLER: https://github.com/mpi-forum/mpi-standard/pull/633 * MPI_DELETE_ERROR_CLASS and friends: https://github.com/mpi-forum/mpi-standard/pull/166 Next Meeting ============ Joint with Session: Monday May 9th 13:00 EST WG only: Monday May 16th, 12:00 EST Best Regards, Aurelien -- Aurelien Bouteiller, Ph.D. Innovative Computing Laboratory; The University of Tennessee; 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) > On May 2, 2022, at 11:39, Ignacio Laguna via mpiwg-ft <mpiwg-ft(a)lists.mpi-forum.org> wrote: > > I have conflict today and can't attend. I will call in next time. > Thanks > Ignacio > > On 5/2/22 8:38 AM, Aurelien Bouteiller via mpiwg-ft wrote: >> As planned we will have a meeting today at noon EST. >> On the agenda are >> 1. changes to fine-grain recovery (aka ulfm v2) from past week plenary >> 2. Quick review of all PR’s touching error handling >> Best, >> Aurelien >>> On Apr 11, 2022, at 10:37, Aurelien Bouteiller <bouteill(a)icl.utk.edu> wrote: >>> >>> Hey all, >>> >>> We will have a joint meeting today with the Sessions WG (connexion info is on their GitHub) at 1pm EST. >>> >>> We will have the solo WG Errors & FT meeting next week at noon EST. We will review all current PRs that touch error handling. >>> >>> >>> In order to clarify our schedule, I have created a new calendar invite with the correct recurrence, you can download it from the WG wiki. https://urldefense.us/v3/__https://github.com/mpiwg-ft/ft-issues/wiki__;!!G… >>> >>> TO SUMMARIZE: THE WG ERRORS & FT MEETINGS WILL BE EVERY 2 WEEKS AT NOON EST, STARTING MON APR. 18. >>> >>> Best Regards, >>> Aurelien >>> >>> -- >>> Aurelien Bouteiller, Ph.D. >>> Innovative Computing Laboratory; The University of Tennessee; >>> 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA >>> +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) >>> >>> >>> >>> >>> >>> >> _______________________________________________ >> mpiwg-ft mailing list >> mpiwg-ft(a)lists.mpi-forum.org >> https://urldefense.us/v3/__https://lists.mpi-forum.org/mailman/listinfo/mpi… > _______________________________________________ > mpiwg-ft mailing list > mpiwg-ft(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-ft

1 0

Re: [mpiwg-ft] WG FT Meeting schedule
by Ignacio Laguna 02 May '22

02 May '22

I have conflict today and can't attend. I will call in next time. Thanks Ignacio On 5/2/22 8:38 AM, Aurelien Bouteiller via mpiwg-ft wrote: > As planned we will have a meeting today at noon EST. > > On the agenda are > 1. changes to fine-grain recovery (aka ulfm v2) from past week plenary > 2. Quick review of all PR’s touching error handling > > > Best, > Aurelien > > >> On Apr 11, 2022, at 10:37, Aurelien Bouteiller <bouteill(a)icl.utk.edu> wrote: >> >> Hey all, >> >> We will have a joint meeting today with the Sessions WG (connexion info is on their GitHub) at 1pm EST. >> >> We will have the solo WG Errors & FT meeting next week at noon EST. We will review all current PRs that touch error handling. >> >> >> In order to clarify our schedule, I have created a new calendar invite with the correct recurrence, you can download it from the WG wiki. https://urldefense.us/v3/__https://github.com/mpiwg-ft/ft-issues/wiki__;!!G… >> >> TO SUMMARIZE: THE WG ERRORS & FT MEETINGS WILL BE EVERY 2 WEEKS AT NOON EST, STARTING MON APR. 18. >> >> Best Regards, >> Aurelien >> >> -- >> Aurelien Bouteiller, Ph.D. >> Innovative Computing Laboratory; The University of Tennessee; >> 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA >> +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) >> >> >> >> >> >> > > _______________________________________________ > mpiwg-ft mailing list > mpiwg-ft(a)lists.mpi-forum.org > https://urldefense.us/v3/__https://lists.mpi-forum.org/mailman/listinfo/mpi… >

1 0

Re: [mpiwg-ft] WG FT Meeting schedule
by Aurelien Bouteiller 02 May '22

02 May '22

As planned we will have a meeting today at noon EST. On the agenda are 1. changes to fine-grain recovery (aka ulfm v2) from past week plenary 2. Quick review of all PR’s touching error handling Best, Aurelien > On Apr 11, 2022, at 10:37, Aurelien Bouteiller <bouteill(a)icl.utk.edu> wrote: > > Hey all, > > We will have a joint meeting today with the Sessions WG (connexion info is on their GitHub) at 1pm EST. > > We will have the solo WG Errors & FT meeting next week at noon EST. We will review all current PRs that touch error handling. > > > In order to clarify our schedule, I have created a new calendar invite with the correct recurrence, you can download it from the WG wiki. https://github.com/mpiwg-ft/ft-issues/wiki > > TO SUMMARIZE: THE WG ERRORS & FT MEETINGS WILL BE EVERY 2 WEEKS AT NOON EST, STARTING MON APR. 18. > > Best Regards, > Aurelien > > -- > Aurelien Bouteiller, Ph.D. > Innovative Computing Laboratory; The University of Tennessee; > 1122 Volunteer Blvd.; Claxton suite 203; 37996, Knoxville, TN, USA > +1 865 974 9308 (p); +1 865 974 8296 (f); Claxton 316 (f2f) > > > > > >

1 0