Add a capability to parse the rankfile to extract node information in place of requiring both hostfile and rankfile for non-RM managed environments. The rankfile is -only- parsed for this IF the hostfile and -host options are not given. Otherwise, those are used to establish allocation info as we did before this commit.
This commit was SVN r21815.
Adds several new mpirun options:
* -bysocket - assign ranks on a node by socket. Effectively load balances the procs assigned to a node across the available sockets. Note that ranks can still be bound to a specific core within the socket, or to the entire socket - the mapping is independent of the binding.
* -bind-to-socket - bind each rank to all the cores on the socket to which they are assigned.
* -bind-to-core - currently the default behavior (maintained from prior default)
* -npersocket N - launch N procs for every socket on a node. Note that this implies we know how many sockets are on a node. Mpirun will determine its local values. These can be overridden by provided values, either via MCA param or in a hostfile
Similar features/options are provided at the board level for multi-board nodes.
Documentation to follow...
This commit was SVN r21791.
the BATCH_PARTITION_ID anymore, use the ras-alps-command.sh script to
figure out the jobs ID to query from ALPS.
Gracefully report errors, update the help file and parse the sysconfig file
This commit was SVN r21772.
variables that are not initialized and are declared in a file that
doesn't export any globally visible function are marked as
non-initialized constants, i.e. uninitialized common symbols. For some
obscure reasons, they get removed from the object files on Mac OS X.
So far I found two solution to this problem. One require the addition
of "-c" to the linker command, the second one (corresponding to this
patch) force them to became a common initialized symbol.
This commit was SVN r21739.
note in VERSION file.
NOTE: the versions will ''always'' be 0:0:0 on the SVN trunk and
developer branches. They will only have meaningful values (starting
with 0:0:0 in 1.3.4) on release branches. Only RM's will modify these
values immediately preceeding a release.
This commit was SVN r21729.
When we read the input buffer, we don't always get a complete printf output - we sometimes end mid stream. We still need to add the suffix and a <CR> to keep the output working right.
This commit was SVN r21706.
"critical" - any error at or above the critical severity will be reported (i.e., only critical errors)
"warning" - any error at or above the warning severity will be reported (i.e., warning and critical errors)
"notice" - pretty much everything will be reported
Default to "critical" to keep down the chatter.
Obviously, only places that call orte_notifier will be affected - all other error reporting (e.g., via opal_output calls) is unaffected.
This commit was SVN r21693.
Add ability to store the RM's jobid string to tag the notifier message so that the sys admin knows what job had the problem.
This commit was SVN r21687.
outside the boundaries of the allocated array.
However, the problem is still there. If we have rmaps file containing only
partial information the num_procs get set to the wrong value (the number of
hosts in the rmaps file instead of the number of processes requested on the
command line).
This commit was SVN r21686.
Remove all architecture references from ORTE and put them back in the modex using modex_send/recv calls.
Hetero operations are now fully supported again. Comm_spawn now works up to the point where it segfaults due to an error in the CID code - which now allows Edgar to dig further! :-)
This commit was SVN r21655.
