3. Customize Attributes¶
Now we will customize the app to work on your cluster. Be sure that you walk through Software Requirements for the given cluster ahead of time.
The main responsibility of the form.yml
file (User Form) located in the root of the app is defining the attributes (their values or HTML form elements) used when generating the batch script.
- Open
~/ondemand/dev/bc_my_center_matlab/form.yml
you should see something that looks like this:
--- cluster: "owens" form: - version - bc_account - bc_num_hours - bc_num_slots - num_cores - node_type - bc_vnc_resolution - bc_email_on_started attributes: num_cores: widget: "number_field" label: "Number of cores" value: 1 help: | Number of cores on node type (4 GB per core unless requesting whole node). Leave blank if requesting full node. min: 0 max: 48 step: 1 id: 'num_cores' bc_num_slots: "1" bc_vnc_resolution: required: true bc_account: label: "Project" help: "You can leave this blank if **not** in multiple projects." node_type: widget: select label: "Node type" help: | - **any** - (*1-28 cores*) Use any available Owens node. This reduces the wait time as there are no node requirements. - **hugemem** - (*48 cores*) Use an Owens node that has 1.5TB of available RAM as well as 48 cores. There are 16 of these nodes on Owens. Requesting hugemem nodes allocates entire nodes. - **vis** - (*1-28 cores*) Use an Owens node that has an [NVIDIA Tesla P100 GPU](http://www.nvidia.com/object/tesla-p100.html) with an X server running in the background. This utilizes the GPU for hardware accelerated 3D visualization. There are 160 of these nodes on Owens. options: - [ "any", "" ] - [ "hugemem", ":hugemem" ] - [ "vis", ":vis:gpus=1" ] version: widget: select label: "MATLAB version" help: "This defines the version of MATLAB you want to load." options: - [ "R2018b", "matlab/r2018b" ] - [ "R2018a", "matlab/r2018a" ] - [ "R2017a", "matlab/r2017a" ] - [ "R2016b", "matlab/r2016b" ] - [ "R2015b", "matlab/r2015b" ]
Note
Multiple form elements / attributes are specific to OSC. We will focus on the ones that are required, and some that may be universally useful.
- We will begin by adding a cluster for the MATLAB app to use. You do this by editing
~/ondemand/dev/bc_my_center_matlab/form.yml
in your favorite editor as such:
# ~/ondemand/dev/bc_my_center_matlab/form.yml --- cluster: "owens" # <---- Change this form: - version # At OSC we provide - bc_account - bc_num_hours - bc_num_slots - num_cores - node_type - bc_vnc_resolution - bc_email_on_started # ...where we replace
owens
with a valid cluster that corresponds to a cluster configuration file located under/etc/ood/config/clusters.d/owens.yml
.Note
The
cluster
attribute is not available to the form and cannot be changed at runtime.
Set common options:
--- cluster: "owens" form: - version # <-- OSC supports multiple versions of MATLAB via LMod - bc_account # <-- the charge account - bc_num_hours # <-- the number of user requested hours that the job run - bc_num_slots # <-- the number of nodes - num_cores # <-- the number of processors per node - node_type # <-- OSC offers different hardware types (any, huge memory, GPU, ...) - bc_vnc_resolution # <-- the resolution settings for the VNC connection - bc_email_on_started # <-- should the system email the user when the job starts
Set attribute values for each form element
num_cores
has a max and min set that are specific to OSC’s Owens cluster; set these values to whatever makes sense for your site.--- attributes: num_cores: widget: "number_field" label: "Number of cores" value: 1 help: | Number of cores on node type (4 GB per core unless requesting whole node). Leave blank if requesting full node. min: 0 max: 48 step: 1 id: 'num_cores'
bc_num_slots
is hardcoded to 1 and is not user editable because OSC does not support MPI work from the Batch Connect app. The value forbc_vnc_resolution
should be copied verbatim.bc_account
is the charge account for the batch work.--- attributes: bc_num_slots: "1" bc_vnc_resolution: required: true bc_account: label: "Project" help: "You can leave this blank if **not** in multiple projects."
node_type
allows users to select which hardware they want to run their work on. In theoptions
mapping the first value is displayed to the user, the second value is made available to any ERB files in~/ondemand/dev/bc_my_center_matlab/template/*
and~/ondemand/dev/bc_my_center_matlab/submit.yml.erb
.version
allows the user to select what version of MATLAB they want to run, and the second value corresponds to OSC’s module names.--- attributes: node_type: widget: select label: "Node type" help: | - **any** - (*1-28 cores*) Use any available Owens node. This reduces the wait time as there are no node requirements. - **hugemem** - (*48 cores*) Use an Owens node that has 1.5TB of available RAM as well as 48 cores. There are 16 of these nodes on Owens. Requesting hugemem nodes allocates entire nodes. - **vis** - (*1-28 cores*) Use an Owens node that has an [NVIDIA Tesla P100 GPU](http://www.nvidia.com/object/tesla-p100.html) with an X server running in the background. This utilizes the GPU for hardware accelerated 3D visualization. There are 160 of these nodes on Owens. options: - [ "any", "" ] - [ "hugemem", ":hugemem" ] - [ "vis", ":vis:gpus=1" ] version: widget: select label: "MATLAB version" help: "This defines the version of MATLAB you want to load." options: - [ "R2018b", "matlab/r2018b" ] - [ "R2018a", "matlab/r2018a" ] - [ "R2017a", "matlab/r2017a" ] - [ "R2016b", "matlab/r2016b" ] - [ "R2015b", "matlab/r2015b" ]
Note
submit.yml.erb
may also besubmit.yml
if there is no requirement to pass form values to the adapter, or perform another server-side task.