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

Re: [mpiwg-ft] help with advice to implementors accompanying MPI_Abort
by Aurelien Bouteiller 25 Jun '19

25 Jun '19

It does not appear to me that any change is necessary. The only part about MPI_COMM_WORLD is in an 'as an example' clause. That being said, substituting MPI_COMM_WORLD with some generic 'communicator comm' would also work. Aurelien On Tue, Jun 25, 2019 at 9:49 AM Pritchard Jr., Howard via mpiwg-ft < mpiwg-ft(a)lists.mpi-forum.org> wrote: > Hello MPI FTer’s, > > > > The Sessions WG could use some help/suggestions about how to adjust the > following advice to implementors that accompanies the definition of > MPI_Abort: > > > > \begin{implementors} > > After aborting a subset of processes, a high quality implementation > should > > be able to provide error handling for communicators, windows, and > files > > involving both aborted and non-aborted processes. As an example, if > the > > user changes the error handler for \const{MPI\_COMM\_WORLD} to > > \const{MPI\_ERRORS\_RETURN} or a custom error handler, when a subset > of > > \const{MPI\_COMM\_WORLD} is aborted, the remaining processes in > > \const{MPI\_COMM\_WORLD} should be able to continue communicating > with each > > other and receive appropriate error codes when attempting > communication > > with an aborted process. > > \end{implementors} > > > > We would like to generalize this advice to implementors to the case where > MPI_COMM_WORLD isn’t a valid communicator, i.e. when an application is > using the Sessions model. > > We think that there would need to be some reworking of the existing text > to cover the sessions use case. Since the FT group has worked quite a bit > on this text, we’d defer to your group for suggestions on how to generalize > this text to cover the sessions use case. > > > > Thanks very much for any help, > > > > Howard > > > > -- > > > > Howard Pritchard > > HPC-ENV > > Los Alamos National Laboratory > > > _______________________________________________ > mpiwg-ft mailing list > mpiwg-ft(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-ft >

1 0

help with advice to implementors accompanying MPI_Abort
by Pritchard Jr., Howard 25 Jun '19

25 Jun '19

Hello MPI FTer’s, The Sessions WG could use some help/suggestions about how to adjust the following advice to implementors that accompanies the definition of MPI_Abort: \begin{implementors} After aborting a subset of processes, a high quality implementation should be able to provide error handling for communicators, windows, and files involving both aborted and non-aborted processes. As an example, if the user changes the error handler for \const{MPI\_COMM\_WORLD} to \const{MPI\_ERRORS\_RETURN} or a custom error handler, when a subset of \const{MPI\_COMM\_WORLD} is aborted, the remaining processes in \const{MPI\_COMM\_WORLD} should be able to continue communicating with each other and receive appropriate error codes when attempting communication with an aborted process. \end{implementors} We would like to generalize this advice to implementors to the case where MPI_COMM_WORLD isn’t a valid communicator, i.e. when an application is using the Sessions model. We think that there would need to be some reworking of the existing text to cover the sessions use case. Since the FT group has worked quite a bit on this text, we’d defer to your group for suggestions on how to generalize this text to cover the sessions use case. Thanks very much for any help, Howard -- Howard Pritchard HPC-ENV Los Alamos National Laboratory

1 0

FTWG Con Call - 2019-06-10
by Wesley Bland 10 Jun '19

10 Jun '19

The Fault Tolerance Working Group’s weekly con call is today at 12:00 PM Eastern. Today's agenda: * Continue fine and course grained recovery discussions (Aurelien and Ignacio, feel free to contribute details of what you’d like to talk about). If there's something else that people would like to discuss, please just send an email to the WG so we can get it on the agenda. Thanks, Wesley ......................................................................................................................................... Join from PC, Mac, Linux, iOS or Android: https://tennessee.zoom.us/j/632356722?pwd=lI4_169CGcewIumekTziMw Password: mpiforum Or iPhone one-tap (US Toll): +16468769923,632356722# or +16699006833,632356722# Or Telephone: Dial: +1 646 876 9923 (US Toll) +1 669 900 6833 (US Toll) Meeting ID: 632 356 722 International numbers available: https://zoom.us/u/6uINe Or an H.323/SIP room system: H.323: 162.255.37.11 (US West) or 162.255.36.11 (US East) Meeting ID: 632 356 722 Password: 364216 SIP: 632356722(a)zoomcrc.com Password: 364216 .........................................................................................................................................

1 0

Re: [mpiwg-ft] Plenary Slides for FTWG Overview
by Wesley Bland 30 May '19

30 May '19

Thanks. I’ve taken care of that. > On May 30, 2019, at 12:18 PM, Laguna Peralta, Ignacio <lagunaperalt1(a)llnl.gov> wrote: > > Hi Wesley, > > Very good slides! They cover well the group discussion topics. > > I have a suggestion on slide 11 ("Reinitialize MPI"): we could add a > line "...initialize..." between the MPI_Init() and the MPI_Reinit() > calls since programmers may want to skip some initialization steps > during recovery by placing the MPI_Reinit() location a bit further in > the main function (but always before any MPI state is created). > > The same concept applies to slide 12 (i.e., adding an "...initialize..." > line after MPI_Init() makes sense). > > Thank you very much for putting this together. > > Ignacio > > > > > On 5/30/19 10:03 AM, Wesley Bland via mpiwg-ft wrote: >> Hi all, >> >> As promised, I’ve added some figures to the slides that I hope to >> present on Friday to give the MPI Forum an update on the work being done >> in the FTWG. Please take a look sometime today if possible before I >> present them. >> >> Thanks, >> Wesley >> >> >> _______________________________________________ >> 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] Plenary Slides for FTWG Overview
by Laguna Peralta, Ignacio 30 May '19

30 May '19

Hi Wesley, Very good slides! They cover well the group discussion topics. I have a suggestion on slide 11 ("Reinitialize MPI"): we could add a line "...initialize..." between the MPI_Init() and the MPI_Reinit() calls since programmers may want to skip some initialization steps during recovery by placing the MPI_Reinit() location a bit further in the main function (but always before any MPI state is created). The same concept applies to slide 12 (i.e., adding an "...initialize..." line after MPI_Init() makes sense). Thank you very much for putting this together. Ignacio On 5/30/19 10:03 AM, Wesley Bland via mpiwg-ft wrote: > Hi all, > > As promised, I’ve added some figures to the slides that I hope to > present on Friday to give the MPI Forum an update on the work being done > in the FTWG. Please take a look sometime today if possible before I > present them. > > Thanks, > Wesley > > > _______________________________________________ > mpiwg-ft mailing list > mpiwg-ft(a)lists.mpi-forum.org > https://lists.mpi-forum.org/mailman/listinfo/mpiwg-ft

1 0

Plenary Slides for FTWG Overview
by Wesley Bland 30 May '19

30 May '19

Hi all, As promised, I’ve added some figures to the slides that I hope to present on Friday to give the MPI Forum an update on the work being done in the FTWG. Please take a look sometime today if possible before I present them. Thanks, Wesley

1 0

FTWG Time Delayed
by Wesley Bland 29 May '19

29 May '19

Hi all, We’re delaying the WG time for this afternoon from 1:30 PM Central US to 2:45 PM Central US. We’ll finish at the same time. Thanks, Wesley

1 0

2019-05-29 WG WebEx Info
by Wesley Bland 29 May '19

29 May '19

Hi all, For the WGs meeting this afternoon, let’s divide the WebExes up like this: Track 1 - Plenaries + FT WG Track 2 - Large Counts WG + Semantic Terms WG The info can be found on the Wiki: https://github.com/mpi-forum/mpi-standard/wiki/MPI-Forum-Webex-Information-… <https://github.com/mpi-forum/mpi-standard/wiki/MPI-Forum-Webex-Information-…> If you don’t have access to that webpage, please email me with your GitHub login information and a quick note saying who you are and where you’re from (just so I can be sure you’re not a bot). Thanks, Wesley

1 0

FTWG Con Call - 2019-05-13
by Wesley Bland 13 May '19

13 May '19

The Fault Tolerance Working Group’s weekly con call is today at 12:00 PM Eastern. Today's agenda: * Finalize any readings / plenaries for the upcoming face to face meeting * Continue discussion of work by Aurelien and Ignacio on roll-forward and roll-back recovery proposals. If there's something else that people would like to discuss, please just send an email to the WG so we can get it on the agenda. Thanks, Wesley ......................................................................................................................................... Join from PC, Mac, Linux, iOS or Android: https://tennessee.zoom.us/j/632356722?pwd=lI4_169CGcewIumekTziMw Password: mpiforum Or iPhone one-tap (US Toll): +16468769923,632356722# or +16699006833,632356722# Or Telephone: Dial: +1 646 876 9923 (US Toll) +1 669 900 6833 (US Toll) Meeting ID: 632 356 722 International numbers available: https://zoom.us/u/6uINe Or an H.323/SIP room system: H.323: 162.255.37.11 (US West) or 162.255.36.11 (US East) Meeting ID: 632 356 722 Password: 364216 SIP: 632356722(a)zoomcrc.com Password: 364216 .........................................................................................................................................

1 0

Re: [mpiwg-ft] FTWG Requests for May 2019 Meeting
by Wesley Bland 29 Apr '19

29 Apr '19

We would actually like to schedule one more plenary to give a short FTWG roadmap update (less technical detail than the talks Aurelien has given). A lot of folks in and out of the forum have asked for this and/or been presenting our work as less than it is, so we’d like to give folks a better idea of our plans. Thanks, Wesley > On Apr 29, 2019, at 11:17 AM, Wesley Bland <work(a)wesbland.com> wrote: > > Hey Martin, > > The FTWG would like to request a half day WG time for a meeting on Tuesday morning if possible. > > We’ll also do no-no readings of: > #119 - Clarify what MPI_SUCCESS guarantees, and what is left undefined when MPI_ERR_XXX happens > #102 - Define error/failure behavior in MPI_INIT/FINALIZE > And Aurelien will read another issue based on the virtual reading. We don’t have an issue/pr for that yet, but I’ll get it to you soon. > > Thanks, > Wesley

1 0