Open MPI logo

Open MPI Development Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Development mailing list

From: Aurelien Bouteiller (bouteill_at_[hidden])
Date: 2007-07-26 16:24:24


Ralph H Castain wrote:
> After some investigation, I'm afraid that I have to report that this - as
> far as I understand what you are doing - may no longer work in Open MPI in
> the future (and I'm pretty sure isn't working in the trunk today except
> [maybe] in the special case of hostfile - haven't verified that).
>
> To ensure we are correctly communicating, let me reiterate what I understand
> you are doing:
>
Correct. Also consider that for my testing I use a batch scheduler that
is not managed by orte right now and provide myself the hostfiles (This
batch scheduler is named OAR and is in use on the grid5000 research
facility in France).

> This was caused by mpirun itself processing its local environment and then
> "pushing" it into the global registry. Keeping everything separated causes a
> bookkeeper's headache and many lines of code that we would like to
> eliminate.
>
>
I see the point. I Agree there is very few benefit at allowing users to
have different local environments on different mpirun instances; while
it should be a real pain to have a clean code managing this. For my sole
usage, the app_context feature you described is a more elegant and
equivalent way of spawning my FT services. I will switch to this right
away.

Still it might be of some use to be able to start different mpirun the
same way you plan comm_spawn to work: sharing the same environment, but
allowing for use of a different hostfile. The use case that comes in
mind is "grid", where different batch schedulers are in use on each
clusters, so you can't gather a single hostfile. This is not a feature I
would fight for, but I can imagine some people might find it useful.

More important for me is the ability to refill the hostfile with fresh
hosts when some of the original ones died. Allocating an huge amount of
spares preventively is just not the correct way to go. On the side I am
not sure that even the best comm_spawn you discussed could be of some
help in this case as I do not want the new nodes to go in a different
COMM_WORLD. Finding a way to update the registry and all the orted to do
so is a much larger issue than simple spawning and I have not been
really thinking about it for now. Maybe we should discuss this issue
separately.

Aurelien
> Please feel free to comment. If this is a big enough issue to a large enough
> audience, then we can try to find a way to solve it (assuming Open MPI's
> community decides to support it).
>
> Ralph
>
>
>
>>>> Next requirement is the ability to add during runtime some nodes to the
>>>> initial pool. Because node may fail (but it is the same with comm_spawn
>>>> basically) , I might need some (lot of) spare nodes to replace failed
>>>> ones. As I do not want to request for twice as many nodes as I need
>>>> (after all, things could just go fine, why should I waste that many
>>>> computing resources for idle spares ?), I definitely want to be able to
>>>> allocate some new nodes to the pool of the already running machines. As
>>>> far as I understand, this is impossible to achieve with the usecase2 and
>>>> quite difficult in usecase1. In my opinion, having the ability to spawn
>>>> on nodes which are not part of the initial hostfile is a key feature
>>>> (and not only for FT purposes).
>>>>
>>>>
>>>>
>>>>
>>> I am looking for more detail into the above issue. What
>>> resource manager are you using?
>>>
>>> Ideally, we would prefer not to support this. Any nodes
>>> that you run on, or hope to run on, would be designated
>>> at the start. For example:
>>>
>>> mpirun -np 1 --host a,b,c,d,e,f,g
>>>
>>> This would cause the one process of the mpi job to start on host a.
>>> Then, the mpi job has available to it the other hosts should it decide
>>> later to start a job on them. However no ORTE daemons would
>>> be started on those nodes until calls to MPI_Comm_spawn
>>> occur. So, the MPI job would not be consuming any resources
>>> until called upon to.
>>>
>> This has actually been the subject of multiple threads on the user list and
>> is considered a critical capability by some users and vendors. I believe
>> there is little problem in allowing those systems that can support it to
>> dynamically add nodes to ORTE via some API into the resource manager. At the
>> moment, none of the RMs support it, but LSF will (and TM at least may)
>> shortly do so, and some of their customers are depending upon it.
>>
>> The problem is that job startup could be delayed for significant time if all
>> hosts must be preallocated. Admittedly, this raises all kinds of issues
>> about how long the job could be stalled waiting for the new hosts. However,
>> as the other somewhat exhaustive threads have discussed, there are computing
>> models that can live with this uncertainty, and RMs that will provide async
>> callbacks to allow the rest of the app to continue working while waiting.
>>
>> Just my $0.00002 - again, this goes back to...are there use-cases and
>> customers to which Open MPI is simply going to say "we won't support that"?
>>
>>
>>> Rolf
>>>
>>>
>>>> I know there have been some extra discussions on this subject.
>>>> Unfortunately it looks like I am not part of the list where it happened.
>>>> I hope those concerns have not been already discussed.
>>>>
>>>> Aurelien
>>>>
>>>> Ralph H Castain wrote:
>>>>
>>>>
>>>>
>>>>> Yo all
>>>>>
>>>>> As you know, I am working on revamping the hostfile functionality to make
>>>>> it
>>>>> work better with managed environments (at the moment, the two are
>>>>> exclusive). The issue that we need to review is how we want the interaction
>>>>> to work, both for the initial launch and for comm_spawn.
>>>>>
>>>>> In talking with Jeff, we boiled it down to two options that I have
>>>>> flow-charted (see attached):
>>>>>
>>>>> Option 1: in this mode, we read any allocated nodes provided by a resource
>>>>> manager (e.g., SLURM). These nodes establish a base pool of nodes that can
>>>>> be used by both the initial launch and any dynamic comm_spawn requests. The
>>>>> hostfile and any -host info is then used to select nodes from within that
>>>>> pool for use with the specific launch. The initial launch would use the
>>>>> -hostfile or -host command line option to provide that info - comm_spawn
>>>>> would use the MPI_Info fields to provide similar info.
>>>>>
>>>>> This mode has the advantage of allowing a user to obtain a large
>>>>> allocation,
>>>>> and then designate hosts within the pool for use by an initial application,
>>>>> and separately designate (via another hostfile or -host spec) another set
>>>>> of
>>>>> those hosts from the pool to support a comm_spawn'd child job.
>>>>>
>>>>> If no resource managed nodes are found, then the hostfile and -host options
>>>>> would provide the list of hosts to be used. Again, comm_spawn'd jobs would
>>>>> be able to specify their own hostfile and -host nodes.
>>>>>
>>>>> The negative to this option is complexity - in the absence of a managed
>>>>> allocation, I either have to deal with hostfile/dash-host allocations in
>>>>> the
>>>>> RAS and then again in RMAPS, or I have "allocation-like" functionality
>>>>> happening in RMAPS.
>>>>>
>>>>>
>>>>> Option 2: in this mode, we read any allocated nodes provided by a resource
>>>>> manager, and then filter those using the command line hostfile and -host
>>>>> options to establish our base pool. Any spawn commands (both the initial
>>>>> one
>>>>> and comm_spawn'd child jobs) would utilize this filtered pool of nodes.
>>>>> Thus, comm_spawn is restricted to using hosts from that initial pool.
>>>>>
>>>>> We could possibly extend this option by only using the hostfile in our
>>>>> initial filter. In other words, let the hostfile downselect the resource
>>>>> manager's allocation for the initial launch. Any -host options on the
>>>>> command line would only apply to the hosts used to launch the initial
>>>>> application. Any comm_spawn would use the hostfile-filtered pool of hosts.
>>>>>
>>>>> The advantage here is simplicity. The disadvantage lies in flexibility for
>>>>> supporting dynamic operations.
>>>>>
>>>>>
>>>>> The major difference between these options really only impacts the initial
>>>>> pool of hosts to be used for launches, both the initial one and any
>>>>> subsequent comm_spawns. Barring any commentary, I will implement option 1
>>>>> as
>>>>> this provides the maximum flexibility.
>>>>>
>>>>> Any thoughts? Other options we should consider?
>>>>>
>>>>> Thanks
>>>>> Ralph
>>>>>
>>>>> ------------------------------------------------------------------------
>>>>>
>>>>> _______________________________________________
>>>>> devel mailing list
>>>>> devel_at_[hidden]
>>>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>>>>
>>>>>
>>>>>
>>>> _______________________________________________
>>>> devel mailing list
>>>> devel_at_[hidden]
>>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> devel mailing list
>>> devel_at_[hidden]
>>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>>
>> _______________________________________________
>> devel mailing list
>> devel_at_[hidden]
>> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>>
>
>
> _______________________________________________
> devel mailing list
> devel_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/devel
>