can only be a TORUS (1).

that happens when it gets to the controller anyway.

The ofed plugin doesn't correctly collect packets_in and packets_out.
The ofed plugin treats them as a running total while the hdf5 profile
plugin thinks they are a difference series.

The attached match makes these data items differences.
bug 430

This specifies the maximum depth the backfill scheduler should go
in any single partition.

Also remove group/other write access to PID file

Remove -C option (implemented improperly)
Support multiple -W options (array instead of single option)
Map -l proc option to job constraint
Fix parsing of wall time with colons (avoid breaking into multiple tokens)
Credit to NCSC

srun is pending.  The step creation would use the old values and possibly
hang srun since the step wouldn't be able to be created in the modified
allocation.

Recent Unix'es support a O_CLOEXEC flag to open(), which avoids the potential race condition between opening and setting close-on-exec with fcntl() (and of course, one syscall less). The attached patch does this for a few cases. There are still many more places where this approach could be used; for sockets there is also the Linux-specific SOCK_CLOEXEC flag.

Prevent vestigial open files for SlurmctldProlog and Epilog

into the future.

Permit the job comment to contain multiple fields

bug 428

This reverts commit db1a86d0.