August 2013 - mpiwg-ft - mailman3.cels.anl.gov

Re: [Mpi3-ft] FTWG Con Call 2013-08-06
by Richard Graham 06 Aug '13

06 Aug '13

Thanks !!! From: mpi3-ft-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Wesley Bland Sent: Tuesday, August 06, 2013 11:00 AM To: MPI 3.0 Fault Tolerance and Dynamic Process Control working Group Subject: Re: [Mpi3-ft] FTWG Con Call 2013-08-06 That was my plan for the last agenda item. We'll take care of it and email you the results. On Tuesday, August 6, 2013 at 9:44 AM, Richard Graham wrote: Wesley, I am not going to be able to make the meeting today, due to a conflict. However, if you could specifically discuss presenting an overview of the proposal (hour or so presentation) at the Madrid meeting this would be good. This is for the benefit of the European’s who will be at the meeting, and do not typically make the US meetings. As I understand it, I will be the only one from the WG in attendance, so don’t mind giving the presentation. Thanks, Rich From: mpi3-ft-bounces(a)lists.mpi-forum.org<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Wesley Bland Sent: Tuesday, August 06, 2013 8:55 AM To: mpi3-ft(a)lists.mpi-forum.org<mailto:[email protected]> Subject: [Mpi3-ft] FTWG Con Call 2013-08-06 Dear WG members, This is a reminder that according to our planning, we are having our regular phone meeting TODAY at 3pm EDT. Sorry about the late posting. Date: August 6, Time: 3pm EDT/New York Dial-in information: 218-339-4600<tel:218-339-4600> Code: 623998# Agenda: * Rationale from face-to-face meeting @ UTK * Discuss plans/timeline for bringing ULFM to the full forum * Plan for Madrid meeting Next Meetings: * August. 20, 2013 * September 3, 2013 * September 17, 2013 * October 1, 2013 _______________________________________________ mpi3-ft mailing list mpi3-ft(a)lists.mpi-forum.org<mailto:[email protected]> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft

1 0

Re: [Mpi3-ft] FTWG Con Call 2013-08-06
by Wesley Bland 06 Aug '13

06 Aug '13

That was my plan for the last agenda item. We'll take care of it and email you the results. On Tuesday, August 6, 2013 at 9:44 AM, Richard Graham wrote: > Wesley, > I am not going to be able to make the meeting today, due to a conflict. However, if you could specifically discuss presenting an overview of the proposal (hour or so presentation) at the Madrid meeting this would be good. This is for the benefit of the European’s who will be at the meeting, and do not typically make the US meetings. As I understand it, I will be the only one from the WG in attendance, so don’t mind giving the presentation. > > Thanks, > Rich > > From: mpi3-ft-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Wesley Bland > Sent: Tuesday, August 06, 2013 8:55 AM > To: mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > Subject: [Mpi3-ft] FTWG Con Call 2013-08-06 > > Dear WG members, > > This is a reminder that according to our planning, we are having our regular phone meeting TODAY at 3pm EDT. Sorry about the late posting. > > Date: August 6, > Time: 3pm EDT/New York > Dial-in information: 218-339-4600 (tel:218-339-4600) > Code: 623998# > > > Agenda: > * Rationale from face-to-face meeting @ UTK > * Discuss plans/timeline for bringing ULFM to the full forum > > * Plan for Madrid meeting > > Next Meetings: > * August. 20, 2013 > * September 3, 2013 > > * September 17, 2013 > > * October 1, 2013 > > > > > > _______________________________________________ > mpi3-ft mailing list > mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft > >

1 0

Re: [Mpi3-ft] FTWG Con Call 2013-08-06
by Richard Graham 06 Aug '13

06 Aug '13

Wesley, I am not going to be able to make the meeting today, due to a conflict. However, if you could specifically discuss presenting an overview of the proposal (hour or so presentation) at the Madrid meeting this would be good. This is for the benefit of the European's who will be at the meeting, and do not typically make the US meetings. As I understand it, I will be the only one from the WG in attendance, so don't mind giving the presentation. Thanks, Rich From: mpi3-ft-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Wesley Bland Sent: Tuesday, August 06, 2013 8:55 AM To: mpi3-ft(a)lists.mpi-forum.org Subject: [Mpi3-ft] FTWG Con Call 2013-08-06 Dear WG members, This is a reminder that according to our planning, we are having our regular phone meeting TODAY at 3pm EDT. Sorry about the late posting. Date: August 6, Time: 3pm EDT/New York Dial-in information: 218-339-4600<tel:218-339-4600> Code: 623998# Agenda: * Rationale from face-to-face meeting @ UTK * Discuss plans/timeline for bringing ULFM to the full forum * Plan for Madrid meeting Next Meetings: * August. 20, 2013 * September 3, 2013 * September 17, 2013 * October 1, 2013

1 0

[Mpi3-ft] ULFM Rationale Slides
by Wesley Bland 06 Aug '13

06 Aug '13

As promised on the call last week, here's some slides discussing the rationale that we developed at the face-to-face meeting @ UTK. We'll discuss this more on the call this afternoon. I'm attaching a Dropbox link this time instead of sending an attachment as those seem to be filtered by the listserv. Some of these slides will be a quick overview for those who might have forgotten the basic ideas of ULFM. ULFM Rationale Slides.pdf (http://db.tt/Zbjrl4bM) Thanks, Wesley

1 0

[Mpi3-ft] FTWG Con Call 2013-08-06
by Wesley Bland 06 Aug '13

06 Aug '13

Dear WG members, This is a reminder that according to our planning, we are having our regular phone meeting TODAY at 3pm EDT. Sorry about the late posting. Date: August 6, Time: 3pm EDT/New York Dial-in information: 218-339-4600 Code: 623998# Agenda: * Rationale from face-to-face meeting @ UTK * Discuss plans/timeline for bringing ULFM to the full forum * Plan for Madrid meeting Next Meetings: * August. 20, 2013 * September 3, 2013 * September 17, 2013 * October 1, 2013

1 0

Re: [Mpi3-ft] Problem with reusing rendezvous memory buffers
by Wesley Bland 02 Aug '13

02 Aug '13

I think you're right. We just need to make sure we're careful. In that case, I withdraw my suggestion. On Thursday, August 1, 2013 at 5:09 AM, George Bosilca wrote: > On Wed, Jul 31, 2013 at 7:28 PM, Pavan Balaji <balaji(a)mcs.anl.gov (mailto:[email protected])> wrote: > > George, > > > > I'm not talking about RMA here. I'm just talking about send/recv. I'm also > > not talking about simultaneous receives posted on the same buffer. They are > > separated by failure notifications, COMM_AGREE, whatever else you need. I > > think Sayantan got the point Wesley mentioned and his solution is correct > > (assuming enough support from the network to do so). Though I'm not > > convinced it's not adding overhead. Sayantan should comment on this. > > > > > If you're not talking RMA then Sayantan is absolutely right, this is a > non issue as any message from a process considered as dead should be > discarded. In fact the current ULFM implementation already take in > account such cases, as they can arrive in any multi-rail situation. > > George. > > > With respect to the K Computer, I don't have a link. This was my > > understanding from what the Fujitsu folks mentioned (and was the reason why > > they didn't want to release their low-level API; since the hardware has no > > protection, anyone could write to anyone's memory). And I'm not trying to > > prove that K computer is not good enough in any way. That was just an > > example. My point was only that you should consider the case that not all > > networks would have such protection capabilities. > > > > -- Pavan > > > > > > On 07/31/2013 11:50 AM, George Bosilca wrote: > > > > > > On Tue, Jul 30, 2013 at 10:47 PM, Pavan Balaji <balaji(a)mcs.anl.gov (mailto:[email protected])> wrote: > > > > > > > > > > > > Hmm. Yes, you are right. Generating different per-process rkeys is an > > > > option on IB. Though that's obviously less scalable than a single rkey > > > > and > > > > hurts performance too because a new rkey has to be generated for each > > > > process. Even more of a reason for FT to have a requested/provided > > > > option > > > > like threads. > > > > > > > > > > > > > > > > I fail to understand the scenario allowing you to reach such a > > > conclusion. You want to have an MPI_RECV with a buffer where multiple > > > senders can do RMA operations. The only way this can be done in the > > > context of the MPI standard is if each of the receives on this > > > particular buffer are using non-contiguous datatypes. Thus, unlike > > > what you suggest in your answer above, this is not hurting performance > > > as you are already in a niche mode (I'm not even talking about the > > > fact that usually non-contiguous datatypes conflicts with RMA > > > operations). Moreover, you suppose that the detection of a dead > > > process and the re-posting of the receive buffer can happen faster > > > than an RMA message cross the network. The only potential case where > > > such a scenario can happen is when multiple paths between the source > > > and the destination exist, and the failure detection happen on one > > > path while the RMA message took another one. This is highly improbable > > > in most cases. > > > > > > There are too many ifs in this scenario to make it plausible. Even if > > > we suppose that all those ifs will be true, as Rich said, this is an > > > issue of [quality of] implementation not MPI standard. A high quality > > > MPI implementation will delay reporting the process failure error on > > > that particular MPI_RECV until all possible RMA from the dead process > > > were either discarded by the network, or written to the memory. > > > > > > > > > > > However, please also think about this problem for other networks that > > > > might > > > > not have such hardware protection capabilities (K Computer comes to > > > > mind). > > > > > > > > > > > > > > > > K computer ? My understanding is that there are such capabilities in > > > the TOFU network. I might be wrong thou, in which case I would > > > definitively appreciate if you can you pinpoint me to a > > > link/documentation that proves your point? > > > > > > > Maybe they cannot provide MPI-specified FT, and that would be fine. > > > > > > > > > Not really, FT can be supported without overhead for the normal > > > execution even for the types of netwrok you mention. The solution I > > > presented above, uses the timeouts of the network layer to ensure no > > > delivery can occur after the error reporting, by delaying the error > > > reporting until all timeout occurred. Trivial to implement, and > > > without impact on the normal execution path. > > > > > > Thanks, > > > George. > > > > > > > > > > > > > > -- Pavan > > > > > > > > > > > > On 07/30/2013 02:59 PM, Sur, Sayantan wrote: > > > > > > > > > > > > > > > Hi Wesley, > > > > > > > > > > Looks like your attachment didn’t make it through. Using IB, one can > > > > > generate rkeys for each sender and just invalidate the key for the > > > > > observed failed process. HW can just drop the “slow” message when it > > > > > arrives. I’m assuming that generating keys should be fast in the future > > > > > given that recently announced HW/firmware has support for on-demand > > > > > registration. In any case, it is not a restriction of IB per se. > > > > > > > > > > Thanks, > > > > > > > > > > Sayantan > > > > > > > > > > *From:*[email protected] (mailto:[email protected]) > > > > > [mailto:[email protected]] *On Behalf Of *Wesley Bland > > > > > *Sent:* Tuesday, July 30, 2013 11:04 AM > > > > > *To:* MPI3-FT Working Group > > > > > *Subject:* [Mpi3-ft] Problem with reusing rendezvous memory buffers > > > > > > > > > > > > > > > Pavan pointed out a problem to me yesterday related to memory buffers > > > > > used with rendezvous protocols. If a process passes a piece of memory to > > > > > the library in an MPI_RECV and the library gives that memory to the > > > > > hardware, where it is pinned, we can get into trouble if one of the > > > > > processes that could write into that memory fails. The problem comes > > > > > from a process sending a slow message and then dying. It is possible > > > > > that the other processes could detect and handle the failure before the > > > > > slow message arrives. Then when the message does arrive, it could > > > > > corrupt the memory without the application having a way to handle this. > > > > > My whiteboard example is attached as an image. > > > > > > > > > > We can't just unmap memory from the NIC when a failure occurs because > > > > > that memory is still being used by another process's message. Some > > > > > hardware supports unmapping memory for specific senders which would > > > > > solve this issue, but some don't, such as InfiniBand, where the memory > > > > > region just has a key and unmapping it removes it for all senders. > > > > > > > > > > This problem doesn't have a good solution (that I've come up with), but > > > > > I did come up with a solution. We would need to introduce another error > > > > > code (something like MPI_ERR_BUFFER_UNUSABLE) that would be able to tell > > > > > the application that the buffer that the library was using is no longer > > > > > usable because it might be corrupted. For some hardware, this wouldn't > > > > > have to be returned, but for hardware where this isn't possible, the > > > > > library could pass this error to the implementation to say that I need a > > > > > new buffer in order to complete this operation. On the sender side, the > > > > > operation would probably complete successfully since to it, the memory > > > > > was still available. That means that there will be some rollback > > > > > necessary, but that's up to the application to figure out. > > > > > > > > > > I know this is an expensive and painful solution, but this is all I've > > > > > come up with so far. Thoughts from the group? > > > > > > > > > > Thanks, > > > > > > > > > > Wesley > > > > > > > > > > > > > > > > > > > > _______________________________________________ > > > > > mpi3-ft mailing list > > > > > mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > > > > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft > > > > > > > > > > > > > > > > > -- > > > > Pavan Balaji > > > > http://www.mcs.anl.gov/~balaji > > > > > > > > _______________________________________________ > > > > mpi3-ft mailing list > > > > mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > > > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft > > > > > > > > > > > > > > > > _______________________________________________ > > > mpi3-ft mailing list > > > mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > > > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft > > > > > > > > > -- > > Pavan Balaji > > http://www.mcs.anl.gov/~balaji > > > > > _______________________________________________ > mpi3-ft mailing list > mpi3-ft(a)lists.mpi-forum.org (mailto:[email protected]) > http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft > >

1 0

Re: [Mpi3-ft] Problem with reusing rendezvous memory buffers
by George Bosilca 01 Aug '13

01 Aug '13

On Wed, Jul 31, 2013 at 7:28 PM, Pavan Balaji <balaji(a)mcs.anl.gov> wrote: > George, > > I'm not talking about RMA here. I'm just talking about send/recv. I'm also > not talking about simultaneous receives posted on the same buffer. They are > separated by failure notifications, COMM_AGREE, whatever else you need. I > think Sayantan got the point Wesley mentioned and his solution is correct > (assuming enough support from the network to do so). Though I'm not > convinced it's not adding overhead. Sayantan should comment on this. If you're not talking RMA then Sayantan is absolutely right, this is a non issue as any message from a process considered as dead should be discarded. In fact the current ULFM implementation already take in account such cases, as they can arrive in any multi-rail situation. George. > With respect to the K Computer, I don't have a link. This was my > understanding from what the Fujitsu folks mentioned (and was the reason why > they didn't want to release their low-level API; since the hardware has no > protection, anyone could write to anyone's memory). And I'm not trying to > prove that K computer is not good enough in any way. That was just an > example. My point was only that you should consider the case that not all > networks would have such protection capabilities. > > -- Pavan > > > On 07/31/2013 11:50 AM, George Bosilca wrote: >> >> On Tue, Jul 30, 2013 at 10:47 PM, Pavan Balaji <balaji(a)mcs.anl.gov> wrote: >>> >>> >>> Hmm. Yes, you are right. Generating different per-process rkeys is an >>> option on IB. Though that's obviously less scalable than a single rkey >>> and >>> hurts performance too because a new rkey has to be generated for each >>> process. Even more of a reason for FT to have a requested/provided >>> option >>> like threads. >> >> >> I fail to understand the scenario allowing you to reach such a >> conclusion. You want to have an MPI_RECV with a buffer where multiple >> senders can do RMA operations. The only way this can be done in the >> context of the MPI standard is if each of the receives on this >> particular buffer are using non-contiguous datatypes. Thus, unlike >> what you suggest in your answer above, this is not hurting performance >> as you are already in a niche mode (I'm not even talking about the >> fact that usually non-contiguous datatypes conflicts with RMA >> operations). Moreover, you suppose that the detection of a dead >> process and the re-posting of the receive buffer can happen faster >> than an RMA message cross the network. The only potential case where >> such a scenario can happen is when multiple paths between the source >> and the destination exist, and the failure detection happen on one >> path while the RMA message took another one. This is highly improbable >> in most cases. >> >> There are too many ifs in this scenario to make it plausible. Even if >> we suppose that all those ifs will be true, as Rich said, this is an >> issue of [quality of] implementation not MPI standard. A high quality >> MPI implementation will delay reporting the process failure error on >> that particular MPI_RECV until all possible RMA from the dead process >> were either discarded by the network, or written to the memory. >> >>> >>> However, please also think about this problem for other networks that >>> might >>> not have such hardware protection capabilities (K Computer comes to >>> mind). >> >> >> K computer ? My understanding is that there are such capabilities in >> the TOFU network. I might be wrong thou, in which case I would >> definitively appreciate if you can you pinpoint me to a >> link/documentation that proves your point? >> >>> Maybe they cannot provide MPI-specified FT, and that would be fine. >> >> >> Not really, FT can be supported without overhead for the normal >> execution even for the types of netwrok you mention. The solution I >> presented above, uses the timeouts of the network layer to ensure no >> delivery can occur after the error reporting, by delaying the error >> reporting until all timeout occurred. Trivial to implement, and >> without impact on the normal execution path. >> >> Thanks, >> George. >> >> >>> >>> -- Pavan >>> >>> >>> On 07/30/2013 02:59 PM, Sur, Sayantan wrote: >>>> >>>> >>>> Hi Wesley, >>>> >>>> Looks like your attachment didnï¿½t make it through. Using IB, one can >>>> generate rkeys for each sender and just invalidate the key for the >>>> observed failed process. HW can just drop the ï¿½slowï¿½ message when it >>>> arrives. Iï¿½m assuming that generating keys should be fast in the future >>>> given that recently announced HW/firmware has support for on-demand >>>> registration. In any case, it is not a restriction of IB per se. >>>> >>>> Thanks, >>>> >>>> Sayantan >>>> >>>> *From:*[email protected] >>>> [mailto:[email protected]] *On Behalf Of *Wesley Bland >>>> *Sent:* Tuesday, July 30, 2013 11:04 AM >>>> *To:* MPI3-FT Working Group >>>> *Subject:* [Mpi3-ft] Problem with reusing rendezvous memory buffers >>>> >>>> >>>> Pavan pointed out a problem to me yesterday related to memory buffers >>>> used with rendezvous protocols. If a process passes a piece of memory to >>>> the library in an MPI_RECV and the library gives that memory to the >>>> hardware, where it is pinned, we can get into trouble if one of the >>>> processes that could write into that memory fails. The problem comes >>>> from a process sending a slow message and then dying. It is possible >>>> that the other processes could detect and handle the failure before the >>>> slow message arrives. Then when the message does arrive, it could >>>> corrupt the memory without the application having a way to handle this. >>>> My whiteboard example is attached as an image. >>>> >>>> We can't just unmap memory from the NIC when a failure occurs because >>>> that memory is still being used by another process's message. Some >>>> hardware supports unmapping memory for specific senders which would >>>> solve this issue, but some don't, such as InfiniBand, where the memory >>>> region just has a key and unmapping it removes it for all senders. >>>> >>>> This problem doesn't have a good solution (that I've come up with), but >>>> I did come up with a solution. We would need to introduce another error >>>> code (something like MPI_ERR_BUFFER_UNUSABLE) that would be able to tell >>>> the application that the buffer that the library was using is no longer >>>> usable because it might be corrupted. For some hardware, this wouldn't >>>> have to be returned, but for hardware where this isn't possible, the >>>> library could pass this error to the implementation to say that I need a >>>> new buffer in order to complete this operation. On the sender side, the >>>> operation would probably complete successfully since to it, the memory >>>> was still available. That means that there will be some rollback >>>> necessary, but that's up to the application to figure out. >>>> >>>> I know this is an expensive and painful solution, but this is all I've >>>> come up with so far. Thoughts from the group? >>>> >>>> Thanks, >>>> >>>> Wesley >>>> >>>> >>>> >>>> _______________________________________________ >>>> mpi3-ft mailing list >>>> mpi3-ft(a)lists.mpi-forum.org >>>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft >>>> >>> >>> -- >>> Pavan Balaji >>> http://www.mcs.anl.gov/~balaji >>> >>> _______________________________________________ >>> mpi3-ft mailing list >>> mpi3-ft(a)lists.mpi-forum.org >>> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft >> >> >> _______________________________________________ >> mpi3-ft mailing list >> mpi3-ft(a)lists.mpi-forum.org >> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft >> > > -- > Pavan Balaji > http://www.mcs.anl.gov/~balaji

1 0

Re: [Mpi3-ft] Problem with reusing rendezvous memory buffers
by Richard Graham 31 Jul '13

31 Jul '13

PSB From: mpi3-ft-bounces(a)lists.mpi-forum.org [mailto:[email protected]] On Behalf Of Wesley Bland Sent: Wednesday, July 31, 2013 1:31 PM To: MPI 3.0 Fault Tolerance and Dynamic Process Control working Group Subject: Re: [Mpi3-ft] Problem with reusing rendezvous memory buffers On Wednesday, July 31, 2013 at 11:50 AM, George Bosilca wrote: On Tue, Jul 30, 2013 at 10:47 PM, Pavan Balaji <balaji(a)mcs.anl.gov<mailto:[email protected]>> wrote: Hmm. Yes, you are right. Generating different per-process rkeys is an option on IB. Though that's obviously less scalable than a single rkey and hurts performance too because a new rkey has to be generated for each process. Even more of a reason for FT to have a requested/provided option like threads. I fail to understand the scenario allowing you to reach such a conclusion. You want to have an MPI_RECV with a buffer where multiple senders can do RMA operations. The only way this can be done in the context of the MPI standard is if each of the receives on this particular buffer are using non-contiguous datatypes. Thus, unlike what you suggest in your answer above, this is not hurting performance as you are already in a niche mode (I'm not even talking about the fact that usually non-contiguous datatypes conflicts with RMA operations). Moreover, you suppose that the detection of a dead process and the re-posting of the receive buffer can happen faster than an RMA message cross the network. The only potential case where such a scenario can happen is when multiple paths between the source and the destination exist, and the failure detection happen on one path while the RMA message took another one. This is highly improbable in most cases. This isn't necessarily about RMA. This can happen with send/receive as well. I don't disagree that this scenario is unlikely, but it could result in a case where the user ends up with bad data and can't do anything about it. [rich] I think that what you are worried about protecting is an OS that does not provide adequate process protection. I don’t know if we want to go there. There are too many ifs in this scenario to make it plausible. Even if we suppose that all those ifs will be true, as Rich said, this is an issue of [quality of] implementation not MPI standard. A high quality MPI implementation will delay reporting the process failure error on that particular MPI_RECV until all possible RMA from the dead process were either discarded by the network, or written to the memory. However, please also think about this problem for other networks that might not have such hardware protection capabilities (K Computer comes to mind). K computer ? My understanding is that there are such capabilities in the TOFU network. I might be wrong thou, in which case I would definitively appreciate if you can you pinpoint me to a link/documentation that proves your point? Maybe they cannot provide MPI-specified FT, and that would be fine. Not really, FT can be supported without overhead for the normal execution even for the types of netwrok you mention. The solution I presented above, uses the timeouts of the network layer to ensure no delivery can occur after the error reporting, by delaying the error reporting until all timeout occurred. Trivial to implement, and without impact on the normal execution path. Thanks, George. -- Pavan On 07/30/2013 02:59 PM, Sur, Sayantan wrote: Hi Wesley, Looks like your attachment didn’t make it through. Using IB, one can generate rkeys for each sender and just invalidate the key for the observed failed process. HW can just drop the “slow” message when it arrives. I’m assuming that generating keys should be fast in the future given that recently announced HW/firmware has support for on-demand registration. In any case, it is not a restriction of IB per se. Thanks, Sayantan *From:*[email protected]<mailto:[email protected]> [mailto:[email protected]] *On Behalf Of *Wesley Bland *Sent:* Tuesday, July 30, 2013 11:04 AM *To:* MPI3-FT Working Group *Subject:* [Mpi3-ft] Problem with reusing rendezvous memory buffers Pavan pointed out a problem to me yesterday related to memory buffers used with rendezvous protocols. If a process passes a piece of memory to the library in an MPI_RECV and the library gives that memory to the hardware, where it is pinned, we can get into trouble if one of the processes that could write into that memory fails. The problem comes from a process sending a slow message and then dying. It is possible that the other processes could detect and handle the failure before the slow message arrives. Then when the message does arrive, it could corrupt the memory without the application having a way to handle this. My whiteboard example is attached as an image. We can't just unmap memory from the NIC when a failure occurs because that memory is still being used by another process's message. Some hardware supports unmapping memory for specific senders which would solve this issue, but some don't, such as InfiniBand, where the memory region just has a key and unmapping it removes it for all senders. This problem doesn't have a good solution (that I've come up with), but I did come up with a solution. We would need to introduce another error code (something like MPI_ERR_BUFFER_UNUSABLE) that would be able to tell the application that the buffer that the library was using is no longer usable because it might be corrupted. For some hardware, this wouldn't have to be returned, but for hardware where this isn't possible, the library could pass this error to the implementation to say that I need a new buffer in order to complete this operation. On the sender side, the operation would probably complete successfully since to it, the memory was still available. That means that there will be some rollback necessary, but that's up to the application to figure out. I know this is an expensive and painful solution, but this is all I've come up with so far. Thoughts from the group? Thanks, Wesley _______________________________________________ mpi3-ft mailing list mpi3-ft(a)lists.mpi-forum.org<mailto:[email protected]> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft -- Pavan Balaji http://www.mcs.anl.gov/~balaji _______________________________________________ mpi3-ft mailing list mpi3-ft(a)lists.mpi-forum.org<mailto:[email protected]> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft _______________________________________________ mpi3-ft mailing list mpi3-ft(a)lists.mpi-forum.org<mailto:[email protected]> http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi3-ft

1 0