on Linux, sending the header and then the message data does severely
impact performances of ptl/tcp :
on the receiver, reading the data can often result in an PMIX_ERR_RESOURCE_BUSY
or PMIX_ERR_WOULD_BLOCK, which ends up degrading performances)
this commit send both header and message data at the same time via writev()
and makes ptl/tcp virtually as efficient as ptl/usock.
Short writev generally occur when the kernel buffer is full, so there is no
point for retrying in this case.
fwiw, no such degradation was observed on OSX.
Refs open-mpi/ompi#2657
Signed-off-by: Gilles Gouaillardet <gilles@rist.or.jp>
The MCA variable code calls the string from value function with a NULL
string to verify values. The verbosity enumerator was not correctly
checking for a non-NULL value before trying to set the string.
Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>
This commit fixes rare race condition that occurs when the process
that is calling `exit(-1)` has delay between fd cleanup and actual
OS-level exit. This may happen if the process has some work to do
`on_exit()`.
**Problem description**:
Consider an application process that has called `exit(nonzero)`, it's
fd's was closed
but it's actual termination at OS level is delayed by some cleanups (eg.
in callbacks registered via `on_exit()`).
Observed sequence of events was the following:
* orted gets stdio disconnection and activating `IOF COMPLETE` state.
* parallel OOB disconnection causes `COMMUNICATION FAILURE` state to be
activated.
* during `COMMUNICATION FAILURE` processing `odls_base_default_wait_local_proc`
is called even though real waitpid wasn't yet called (code mentions that
waitpid might not be called for unspecified reason). Because of that real exit
code is unknown and set to 0. `odls_base_default_wait_local_proc` callback sees
`IOF COMPLETE` flag and in conjunction with 0-exit-code it activates
`WAITPID FIRED` state.
* processing of `WAITPID FIRED` leads to `NORMALLY TERMINATED` to be
activated.
* `NORMALLY TERMINATED` state in particular leads `ORTE_PROC_FLAG_ALIVE` flag
for this proc to be dropped.
* when application process finally exits and `wait_signal_callback` is
launched. It sets real exit code and calls `odls_base_default_wait_local_proc`
again but at this time since the process has `ORTE_PROC_FLAG_ALIVE` flag
dropped `WAITPID FIRED` state is activated (instead of `EXITED WITH NON-ZERO`)
leading to a hang that was observed.
Signed-off-by: Artem Polyakov <artpol84@gmail.com>
Samples are taken after MPI_Init, and then again after MPI_Barrier. This allows the user to see memory consumption caused by add_procs, as well as any modex contribution from forming connections if pmix_base_async_modex is given.
Using the probe simply involves executing it via mpirun, with however many copies you want per node. Example:
$ mpirun -npernode 2 ./mpi_memprobe
Sampling memory usage after MPI_Init
Data for node rhc001
Daemon: 12.483398
Client: 6.514648
Data for node rhc002
Daemon: 11.865234
Client: 4.643555
Sampling memory usage after MPI_Barrier
Data for node rhc001
Daemon: 12.520508
Client: 6.576660
Data for node rhc002
Daemon: 11.879883
Client: 4.703125
Note that the client value on node rhc001 is larger - this is where rank=0 is housed, and apparently it gets a larger footprint for some reason.
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Revamp the event notification integration to rely on the PMIx event chaining and remove the duplicate chaining in OPAL. This ensures we get system-level events that target non-default handlers.
Restore the hostname entries for MPI-level error messages, but provide an MCA param (orte_hostname_cutoff) to remove them for large clusters where the memory footprint is problematic. Set the default at 1000 nodes in the job (not the allocation).
Begin first cut at memory profiler
Some minor cleanups of memprobe
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Plug a minor memory leak. Tell the PMIx server not to create a dstore memory region for the daemon job as there is nobody to share it with.
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Protect users of hwloc membind functions
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Update PMIx to include NULL string protection
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Update to PMIx master to include key overwrite protection
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
There are only five places in the non-daemon code paths where opal_hwloc_topology is currently referenced:
* shared memory BTLs (sm, smcuda). I have added a code path to those components that uses the location string
instead of the topology itself, if available, thus avoiding instantiating the topology
* openib BTL. This uses the distance matrix. At present, I haven't developed a method
for replacing that reference. Thus, this component will instantiate the topology
* usnic BTL. Uses the distance matrix.
* treematch TOPO component. Does some complex tree-based algorithm, so it will instantiate
the topology
* ess base functions. If a process is direct launched and not bound at launch, this
code attempts to bind it. Thus, procs in this scenario will instantiate the
topology
Note that instantiating the topology on complex chips such as KNL can consume
megabytes of memory.
Fix pernode binding policy
Properly handle the unbound case
Correct pointer usage
Do not free static error messages!
Signed-off-by: Ralph Castain <rhc@open-mpi.org>
Looks like we missed one place where we needed to swap OMPI_UNIQUE for
OMPI_FLAGS_UNIQUE. Thanks to Phil Tooley (@Telemin) for reporting the
issue.
Fixes#2635.
Signed-off-by: Jeff Squyres <jsquyres@cisco.com>
