With slurm-15.08.10, we're seeing occasional segfaults in slurmstepd. The logs point to the following line: slurm-15.08.10/src/slurmd/slurmstepd/mgr.c:2612

On that line, _get_primary_group() is accessing the results of getpwnam_r():
    *gid = pwd0->pw_gid;

If getpwnam_r() cannot find a matching password record, it will set the result (pwd0) to NULL, but still return 0. When the pointer is accessed, it will cause a segfault.

Checking the result variable (pwd0) to determine success should fix the issue.

Do not attempt to power down a node which has never responded if the
    slurmctld daemon restarts without state.
bug 2698

requesting the RUNNING state.

Backport of commit cca1616b from 16.05

Some systems return EWOULDBLOCK rather than EAGAIN on recv() failure
This is an enhancement to commit af47b4b2

See bug 2672 for details

of Slurm.

Compiler errors out preventing these 13 from running without fixing the implied
int type for main.

though we decided to push that fix to 16.05 since this has been broken for
a while and no one has complained.

Avoid error message of "Requested cpu_bind option requires entire node to
    be allocated; disabling affinity" being generated in some cases where
    task/affinity and task/cgroup plugins used together.

restart of the slurmctld.

On some systems the char_to_val was not being put into the plugin,
resulting in the following error:
slurmstepd: [23.0]: symbol lookup error: /home/jette/SLURM/install_smd/lib/slurm/task_cgroup.so: undefined symbol: char_to_val
The problem was fixed by declaring the function "static". The
function was name was also updated with a leading "_" to indicate
the function is local to that module.

Otherwise miscalculated limit will lead to job cancellation even when well
inside the allocated amount.

Bug 2660.

in the slurmdbd segfaulting.

Bug 2656

burst_buffer/cray - Don't call Datawarp "paths" function if script includes
    only create or destroy of persistent burst buffer. Some versions of Datawarp
    software return an error for such scripts, causing the job to be held.
bug 2624

No change in any logic or definitions

If a job fails stage in, set its reason to BurstBufferOperation
with a string describing what happened. Previously the reason was
set to AdminHeld on stage-in failure.

that wasn't set up correctly.