Re: [Dmtcp-forum] Regarding multiple checkpoint/restart

[Devarshi G. <dgh...@um...>]

Hi Jason,

Thanks. I was able to checkpoint from the program.

On Thu, Jul 8, 2010 at 12:23 PM, Jason Ansel <ja...@cs...> wrote:

> Also make sure you run:
> dmtcp_checkpoint cmd
> instead of
> cmd
>
> Thats the most likely cause for dmtcpIsEnabled() being false.
>
> --Jason
>
> On Thu, Jul 8, 2010 at 9:18 AM, Kapil Arya <ka...@cc...> wrote:
> > Hello Devarshi,
> > Sorry for the delayed response. There is an example on using DMTCP API in
> > dmtcp-root/test/dmtcpaware1.c. You can look into this file to get the
> idea.
> > Thanks,
> > -Kapil
> >
> > On Tue, Jul 6, 2010 at 3:34 PM, Devarshi Ghoshal <dgh...@um...>
> > wrote:
> >>
> >> Hi,
> >> I am running my C program using dmtcp_checkpoint(), but
> >> still dmtcpIsEnabled() returns false.
> >> Can you please send an example implementing the programming interface.
> >>
> >> On Mon, Jul 5, 2010 at 8:31 PM, Kapil Arya <ka...@cc...> wrote:
> >>>
> >>> Hello Devarshi,
> >>>
> >>> In most cases, the restarted process will get a new PID. DMTCP
> >>> virtualizes the PIDs and TIDs and hence there is no need for the
> process to
> >>> get the same pid. There is a rare possible coincidence when the
> restarted
> >>> process can get the same PID -- when the PIDs get wrapped around and
> the
> >>> next available PID to be allocated to the restarted process is the same
> as
> >>> its original one; but as I mentioned, its a rare coincidence.
> >>>
> >>> The restarted and the original process can definitely exist at the same
> >>> time, but you will need to start a *new* dmtcp_coordinator for the
> restarted
> >>> process. The reason for this is that DMTCP gives each process a
> unique-id
> >>> which remains the same through the checkpoint-restart cycles and thus
> if
> >>> there are two processes with the same unique-id (parent and restarted)
> and
> >>> they both connect to the same dmtcp_coordinator, it won't work.
> >>>
> >>> However, if you start an extra dmtcp_coordinator, then both processes
> >>> will be talking to two different dmtcp_coordinators and hence there
> won't be
> >>> any issues. Of-course, you need to take care of any conflict of
> resources
> >>> between the restarted and the original process.
> >>>
> >>> Please let us know if you have any more questions.
> >>>
> >>> Thanks,
> >>> -Kapil
> >>>
> >>> On Mon, Jul 5, 2010 at 3:33 PM, Devarshi Ghoshal <
> dgh...@um...>
> >>> wrote:
> >>>>
> >>>> Hi,
> >>>> I am working with DMTCP for one of my projects and have a few
> questions
> >>>> related to it:
> >>>> i) Is there any way that the restarted process gets a new PID?
> >>>> ii) If that's the case, can the restarted process and the parent
> process
> >>>> be executed at the same time?
> >>>> Actually I tried to execute the main process and checkpointed segment
> at
> >>>> the same time but the coordinator doesn't allow because of the
> existing
> >>>> peers. Am I doing something wrong or is this not supported by DMTCP?
> >>>>
> >>>> --
> >>>> Thanks & regards,
> >>>> Devarshi
> >>>>
> >>>>
> >>>>
> ------------------------------------------------------------------------------
> >>>> This SF.net email is sponsored by Sprint
> >>>> What will you do first with EVO, the first 4G phone?
> >>>> Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> >>>> _______________________________________________
> >>>> Dmtcp-forum mailing list
> >>>> Dmt...@li...
> >>>> https://lists.sourceforge.net/lists/listinfo/dmtcp-forum
> >>>>
> >>>
> >>
> >>
> >>
> >> --
> >> Thanks & regards,
> >> Devarshi
> >
> >
> >
> ------------------------------------------------------------------------------
> > This SF.net email is sponsored by Sprint
> > What will you do first with EVO, the first 4G phone?
> > Visit sprint.com/first -- http://p.sf.net/sfu/sprint-com-first
> > _______________________________________________
> > Dmtcp-forum mailing list
> > Dmt...@li...
> > https://lists.sourceforge.net/lists/listinfo/dmtcp-forum
> >
> >
>



-- 
Thanks & regards,
Devarshi