Bug 11318

This reverts commit 69d5d94b.

submit_dir != working dir

In 7092, the customer was seeing the interactive srun set
SLURM_SUBMIT_DIR to the cwd in the interactive srun after -D had been
applied -- which is wrong behavior. This will be fixed in 21.08.

Bug 11407

Bug 11447

Bug 11396

Bug 11363

Signed-off-by: Tim Wickberg <tim@schedmd.com>

Bug 11093

It can happen that child proc fails at some point, and returns some exit
code different from 0. But parent didn't check this properly, and
finished returning w/o error.

Now, if child fails, parent too. This spanws back the error and we know
this job will have a problem.

Bug 11093

container_p_restore get now the list of jobs running from the spool dir
with stepd_available.

Then, it iterates over basepath entries and, for those which seems to
have been a mount point (has .ns file), tries to mount it again.

If it succeeds (it must), and if for this mount point the job is dead,
it releases resources and tries to delete files. Remember the removal
can fail if a resource is leaked. These would be fixed if slurmd starts
after HW reboot (no kernel leaks).

Bug 11093

Then, make _rm_data call in _create_ns keep old behavior, as
here it will only be removed if something fails in creation, and no
previous NS leak is possible. So force removal, or fail.

But for _delete_ns, it could be called at job's end after slurmd
got killed and restarted, thus having leaked the NS. Even though,
slurmd can recreate the NS and mount it in the same place, the .ns
file can't be removed at the end of the job because EBUSY.

Bug 11093

In preparation for subsequent changes.

No functional change.

Bug 11093

This flags checks if we are going to allow the NS creation over a
previously used mount point.

Right now, no change in behavior. Since the only call to _create_ns has
remount to false, thus falling to the old behavior.

In preparation for subsequent changes.

Bug 11093

In preparation for subsequent changes.

No functional change.

Bug 11093

We shouldn't umount the NSs until the step FD is closed.

In preparation for next commit as well.

Bug 11093

Now, if umount2 for basepath fails, it still frees everything up.

In preparation for subsequent changes as well.

Bug 11093

Bug 11257

assoc_mgr_user_list should never be NULL here if we're enforcing associations, however, the issue of going into _get_assoc_mgr_user_list will only happen if locks were already acquired.

This part will be revisited in master for code simplification.

Bug 11257

After 6db0aca5 we acquire assoc_mgr_lock for QoS outside of the loop,
in assoc_mgr_get_admin_level we need to acquire assoc_mgr_lock for user.
This is technically OK, but we want to avoid nested use of
assoc_mgr_lock, which is checked by _store_locks assertion.

validate_operator is often called in loops while
assoc_mgr_get_admin_level is a heavy operation. This commit changes the
functions to lighter versions geting slurmdb_user_rec_t instead of uid.

Bug 11257

Add versions of part_is_visible, validate_operator that get
slurmdb_user_rec_t* instead of uid. In case of execution in a loop it's
more profitable from performance stand point to get the slurmdb_user_rec
once and pass it to functions just performing logical checks.

Bug 11257

This problem has been around a while, but the compiler (gcc 10.2.0)
just now noticed it.

Bug 10407

Bug 11334.

Bug 11245

Bug 11301

Fix regression from a2678ab5.

The original intent of a2678ab5 was to set port in node structure
from the SlurmdPort when nothing was given in NodeName line.
Unfortunately it changed the behavior for multiple nodes defined in one
line with only a single Port value, like: NodeName=test[01-02] Port=3022
will set test01 on 3022, but test02 on SlurmdPort.

Bug 11384

It's short for '--error'.

Bug 11393.

Also update quickstart to list all client commands

Bug 10835

Bug 10439

Signed-off-by: Scott Jackson <scottmo@schedmd.com>

Also remove unnecessary user_name variable.

Bug 10439

Signed-off-by: Scott Jackson <scottmo@schedmd.com>

This avoids intermittent race condition failures and the time to
run the test was significantly reduced.

Bug 10439

Created wait_for_command that repeats a command until successful or until a
specified condition is met. wait_for_command_match is now implemented in
terms of the new wait_for_command. Tests using the former wait_for_command
were refactored as necessary.

Bug 10439

Continued from 22b995c0

Bug 9405

Bug 11341

Instead of using validate_slurm_user() in auth/jwt implement the check
directly. The function is an external in slurmctld, and calling it from
slurmdbd segfaults due to the missing symbol.

Bug 11350

Fix typo for 'priority'
ConstrainRAMSpace can't be used to enforce swappiness
Improve wording in ConstrainSwapSpace

Bug 11307

Bug 9363

Coverity CID 220675

Bug 11262

Signed-off-by: Tim Wickberg <tim@schedmd.com>

Continuation of commit ea667809

Bug 11262

Signed-off-by: Tim Wickberg <tim@schedmd.com>

UnkillableStepProgram exposes environment variables to the script.
Those variables are now documented.

Bug 11231