Attaching to an MPICH Job

Debugging Memory Problems with MemoryScape™ : Chapter 4 Creating Programs for Memory Debugging : Setting Up MPI Debugging Sessions : Debugging MPICH Applications : Attaching to an MPICH Job

You can attach to an MPICH application even if it was not started under MemoryScape’s control. These processes, however, must have previously been linked with the MemoryScape agent. See “Linking Your Application with the Agent”.

To attach to an MPICH application:

1. Start MemoryScape.

Select Attach to running program from the Add Programs to Your MemoryScape Session screen. You are now shown processes that are not yet owned.

2. Attach to the first MPICH process in your workstation cluster by selecting it, then clicking Next.

3. On an IBM SP with the ch_mpi device, attach to the poe process that started your job. For details, see “Starting MemoryScape on a PE Program”.

Normally, the first MPICH process is the highest process with the correct program name in the process list. Other instances of the same executable can be:

— The p4 listener processes if MPICH was configured with ch_p4.

— Additional slave processes if MPICH was configured with ch_shmem or ch_lfshmem.

— Additional slave processes if MPICH was configured with ch_p4 and has a file that places multiple processes on the same computer.

4. After you attach to your program’s processes, you are prompted to also attach to slave MPICH processes. To do so, press Return or choose Yes. If you choose Yes, MemoryScape starts the server processes and acquires all MPICH processes.

In some situations, the processes you expect to see might not exist (for example, they may crash or exit). MemoryScape acquires all the processes it can and then warns you if it can not attach to some of them. If you attempt to dive into a process that no longer exists (for example, using a message queue display), MemoryScape reports hat the process no longer exists.