Use preferred capitalization of "Slurm"

https://slurm.schedmd.com/faq.html#acronym

faqs.rst

@@ -26,15 +26,15 @@ What does it mean for a cluster to deploy Flux?
 
 Most of the time when someone talks about Flux, they will be describing the combined install
 of several projects here that manifest in a full cluster to submit workflows.
-This cluster is comparable to other job managers like SLURM or SGE in that it can be installed
+This cluster is comparable to other job managers like Slurm or SGE in that it can be installed
 as the main workload manager for a site.
 
 Where does Flux work?
 =====================
 
 You likely are associating Flux with high performance computing in that it is comparable
 to other job managers. However, Flux has a unique ability to nest, meaning you (as a user) could
-launch a Flux Instance under a slurm allocation, for example. Along with scheduler nesting,
+launch a Flux Instance under a Slurm allocation, for example. Along with scheduler nesting,
 you can easily demo Flux in a container, or even used in Kubernetes with the
 `Flux Operator <https://flux-framework.org/flux-operator>`_. We have a vision for Flux to
 allow for converged computing, or making it easy to move between traditional HPC and cloud.
@@ -150,7 +150,7 @@ Resources Questions
 Why is Flux ignoring my Nvidia GPUs?
 ====================================
 
-When Flux is launched via a foreign resource manager like SLURM or LSF,
+When Flux is launched via a foreign resource manager like Slurm or LSF,
 it must discover available resources from scratch using
 `hwloc <https://www.open-mpi.org/projects/hwloc/>`_.  To print a resource
 summary, run:
@@ -325,7 +325,7 @@ may help improve efficiency and throughput:
   with ``flux resource drain 0``.
 
 Since Flux can be launched as a parallel job within foreign resource managers
-like SLURM and LSF, your efforts to develop an efficient batch or workflow
+like Slurm and LSF, your efforts to develop an efficient batch or workflow
 management script that runs within a Flux instance can be portable to those
 systems.
 
@@ -337,10 +337,10 @@ How do I run job steps?
 
 A Flux batch job or allocation started with ``flux batch`` or
 ``flux alloc`` is actually a full featured Flux instance run as a job
-within the enclosing Flux instance.  Unlike SLURM, Flux does not have a
+within the enclosing Flux instance.  Unlike Slurm, Flux does not have a
 separate concept like *steps* for work run in a Flux subinstance--we just have
 *jobs*.  That said, a batch script in Flux may contain multiple
-``flux run`` commands just as a SLURM batch script may contain multiple
+``flux run`` commands just as a Slurm batch script may contain multiple
 ``srun`` commands.
 
 Despite there being only one type of *job* in Flux, running a series of jobs
@@ -665,8 +665,8 @@ with-pm=slurm
 
 .. note::
    It appears that ``--with-pm=slurm`` is not required to run MPI programs
-   under SLURM, although it is unclear whether there is a performance impact
-   under SLURM when this option is omitted.
+   under Slurm, although it is unclear whether there is a performance impact
+   under Slurm when this option is omitted.
 
 .. _mpi_init_problems:
 

guides/learning_guide.rst

@@ -459,7 +459,7 @@ management allows this class of workflows to run many modeling application tasks
 each with a different scenario, efficiently at once on a large resource allocation.
 With strict deadline requirements, these workflows must also portably leverage
 computing resources from multiple institutions include the world's most powerful
-supercomputers at ORNL, LLNL, and NERSC. Existing software tools (e.g., SLURM
+supercomputers at ORNL, LLNL, and NERSC. Existing software tools (e.g., Slurm
 and IBM LSF fall short of meeting all of the project's requirements - high job
 throughput, co-scheduling ability, and portability between different HPC systems. Luc
 Peterson, software architect of a main tool used for the NVBL team reported, "With
@@ -613,7 +613,7 @@ and launching the simulations in a general manner becomes difficult.
 The first version of the Merlin MPI parallel job launcher used a simple Python-
 based subprocess call to map a set of MPI parameters (e.g., number of nodes and
 CPU cores) onto the commands needed for launching under the system workload
-manager such as SLURM or IBM LSF. A maintenance issue arose when each new
+manager such as Slurm or IBM LSF. A maintenance issue arose when each new
 workload manager required a set of runtime parameters that do not map 1:1
 between the various launch systems. Moreover, IBM LSF does not handle nested
 launches where there is one subprocess call for the allocation and a subsequent
@@ -647,7 +647,7 @@ Figures 3a and 3b provide simple examples. We first note that the first lines
 these scripts are nearly identical to that of traditional solutions. Exploiting this
 property, Flux can easily be adapted and used with a computing center's existing
 system workload managers with just a few keystrokes. For example, Figure 7 shows
-how Flux can enable the complex workflow in Figure 3b under SLURM, a traditional
+how Flux can enable the complex workflow in Figure 3b under Slurm, a traditional
 HPC workload manager, with just a two-line change in commands. This feature
 has proven to be critical in helping combat time-critical workflow problems such as
 as in COVID-19 research. Thanks to this easy-to-adapt feature, Flux has enabled
@@ -661,7 +661,7 @@ system-instance technology of Flux.
         srun -N256 -n256 flux start <SCRIPT in figure 3b>
 
 The above is Figure 7, and shows that enabling Flux under another workload manager like
-SLURM is as simple as executing this two-line command.
+Slurm is as simple as executing this two-line command.
 
 
 Flux's adaptability to different use cases, along with being open source, has
@@ -947,7 +947,7 @@ to make scheduling descisions. This model is simple and effective for moderate-s
 clusters, making it the state of the practice in most cloud-based and HPC centers
 today. Cloud workload managers (or often called container orchestration solutions)
 such as Swarm and Kubernetes and HPC workload managers such as
-SLURM, MOAB, IBM LSF, and PBSPro (OpenPBS) are centralized.
+Slurm, MOAB, IBM LSF, and PBSPro (OpenPBS) are centralized.
 The cloud products—Kubernetes in particular—can achieve high job throughput,
 but they are incapable of efficient batch job scheduling and rely on overly simplistic
 resource models, resulting in poor performance for HPC workloads. On the other
@@ -969,7 +969,7 @@ like Mesos and YARN as well as the grid solutions like Globus and
 HTCondor. Efforts to achieve better scalability in HPC have resulted in this
 model's implementation at some HPC centers. For example, in the past LLNL
 managed multiple clusters with a limited hierarchical workload manager that used
-the MOAB grid meta-scheduler on top of several SLURM workload managers, each
+the MOAB grid meta-scheduler on top of several Slurm workload managers, each
 of which managed a single cluster in the facility. While this type of solution
 increases scalability over centralized scheduling, it is ultimately limited by its shallow
 (and therefore inflexible) hierarchy and the capabilities of the scheduling frameworks
@@ -1052,7 +1052,7 @@ will enable it to be the primary system workload manager on exascale-computing-
 class supercomputers by 2023, is actively being pursued as additional features and
 performance/scalability tuning, commensurate with the capabilities of then the
 world's fastest supercomputers, are required. It is important to note that existing
-workload managers in HPC (such as SLURM or LSF) have been developed and
+workload managers in HPC (such as Slurm or LSF) have been developed and
 stabilized over a span of decades, until many HPC sites across the world adopted
 and deployed them, and they continue to add features. Similar expectations apply to
 the Flux framework as development effort and feature enhancements continue.

includes/comparison-table.rst

@@ -9,7 +9,7 @@
 
    * - Features
      - Flux
-     - SLURM
+     - Slurm
      - PBSPro (OpenPBS)
      - LSF
      - MOAB 
@@ -46,7 +46,7 @@
 
    * - Features
      - Flux
-     - SLURM
+     - Slurm
      - PBSPro (OpenPBS)
      - LSF
      - MOAB 
@@ -163,7 +163,7 @@
 
    * - Features
      - Flux
-     - SLURM
+     - Slurm
      - PBSPro (OpenPBS)
      - LSF
      - MOAB 
@@ -182,7 +182,7 @@
      - yes
      - yes
    * - Support for nesting within foreign resource manager
-     - yes (slurm, lsf, ...)
+     - yes (Slurm, lsf, ...)
      - n/a
      - n/a
      - n/a

jobs/batch.rst

@@ -107,11 +107,11 @@ Checking the output file of one of the batch job:
 
 
 ------------------------------------
-Launching Flux in SLURM's Batch Mode
+Launching Flux in Slurm's Batch Mode
 ------------------------------------
 
 Users may want to script the above procedures within a script
-to submit to another resource manager such SLURM.
+to submit to another resource manager such Slurm.
 
 An example sbatch script:
 

jobs/debugging.rst

@@ -77,7 +77,7 @@ to ``totalview``. For example,
 
   $ totalview -s tvdrc --args flux run -N 2 -n 2 ./mpi-program
 
-Notice that it is designed to support not only Flux but also SLURM's
+Notice that it is designed to support not only Flux but also Slurm's
 srun and IBM JSM's jsrun commands. The ``regex`` syntax of
 ``exec_handling`` within TotalView can be found in `TotalView user guide`_.
 

spell.en.pws

@@ -237,7 +237,7 @@ PROCID
 rc
 SIGKILL
 sizesp
-SLURM
+Slurm
 stepid
 str
 Unpublish

tutorials/commands/ssh-across-clusters.rst

@@ -17,11 +17,11 @@ and run ``flux start`` via our job manager. Here we might be on a login node:
 
 .. code-block:: console
 
-    # slurm specific
+    # Slurm specific
     $ salloc -N4 --exclusive
     $ srun -N4 -n4 --pty --mpibind=off flux start
 
-And then we get our allocation! You might adapt this command to be more specific to your resource manager. E.g., slurm uses srun.
+And then we get our allocation! You might adapt this command to be more specific to your resource manager. E.g., Slurm uses srun.
 After you run ``flux start``, you are inside of a Flux instance on your allocation!
 Let's run a simple job on our allocation. This first example will ask to see the hostnames of your nodes: