HPCwiki - Recent changes [en]

Reporting Usage

Honfi001 — Thu, 08 Oct 2026 13:48:49 GMT

← Older revision		Revision as of 13:48, 8 October 2026
Line 39:		Line 39:
	== Cost alerts ==		== Cost alerts ==

	You can have Anunna email you and e.g. your supervisors when your bill passes a threshold. Configure, change, or remove the alert with:		You can have Anunna email you and/or e.g. your supervisors when your bill passes a threshold. Configure, change, or remove the alert with:

	<syntaxhighlight lang="bash">module load anunna		<syntaxhighlight lang="bash">module load anunna

2026 Course dates

Honfi001 — Thu, 08 Oct 2026 08:35:12 GMT

← Older revision		Revision as of 08:35, 8 October 2026
(3 intermediate revisions by the same user not shown)
Line 13:		Line 13:
	\|08:30 – 12:00		\|08:30 – 12:00
	\|12-05-26		\|12-05-26
	\|12.30 - 17.00		\|12:30 – 17:00
	\|20-10-26		\|20-10-26
	\|08.30 – 12.00		\|08:30 – 12:00
	\|-		\|-
	\|[[HPC Basic]]		\|[[HPC Basic]]
Line 21:		Line 21:
	\|12:30 – 17:00		\|12:30 – 17:00
	\|09-06-26		\|09-06-26
	\|12.30 - 17.00		\|12:30 – 17:00
	\|10-11-26		\|10-11-26
	\|~~08.30~~ – ~~12.~~00		\|09:00 – 16:00
	\|-		\|-
	\|[[HPC Advanced]]		\|[[HPC Advanced]]
Line 29:		Line 29:
	\|12:30 – 17:00		\|12:30 – 17:00
	\|23-06-26		\|23-06-26
	\|08.30 – 12.00		\|08:30 – 12:00
	\|24-11-26		\|24-11-26
	\|~~12.30 - 17.~~00		\|09:00 – 16:00
	\|}		\|}
	Registration can be done by:		Registration can be done by:

	* sending an e-mail to [mailto:it-workshops@wur.nl it-workshops@wur.nl]		# sending an e-mail to [mailto:it-workshops@wur.nl it-workshops@wur.nl]
	* Mentioning the course(s) you would like to attend		# Mentioning the course(s) you would like to attend
	* Mentioning from which WUR group/team or company you are		# Mentioning from which WUR group/team or company you are
	* Mentioning (if known) if your group is already committed to using Anunna		# Mentioning (if known) if your group is already committed to using Anunna

	As soon as your registration is processed, you will receive a notification and an Invitation will popup in your calendar.		As soon as your registration is processed, you will receive a notification and an Invitation will popup in your calendar.

Workshops

Honfi001 — Thu, 08 Oct 2026 08:31:21 GMT

← Older revision		Revision as of 08:31, 8 October 2026
Line 4:		Line 4:

	* [[Linux Basic]] — Linux command-line fundamentals; the prerequisite for the HPC and container courses.		* [[Linux Basic]] — Linux command-line fundamentals; the prerequisite for the HPC and container courses.
	* [[HPC Basic]] — ~~getting started with the SLURM job scheduler~~.		* [[HPC Basic]] — Introduction to supercomputing principles and basic usage of Anunna.
	* [[HPC Advanced]] — ~~techniques for~~ more ~~demanding and unusual jobs~~.		* [[HPC Advanced]] — Deeper insight into more complex supercomputing workflows at Anunna.

	== Dates and registration ==		== Dates and registration ==

Ollama

Haars001 — Fri, 02 Oct 2026 08:47:38 GMT

Created page with "== How to use ollama on the HPC == On the HPC we have <code>ollama</code> available through modules, in the GPU bucket. To load the module, do this:<syntaxhighlight lang="bash"> ml GPU ollama </syntaxhighlight>Once the module is loaded, <code>ollama</code> is available on your path:<syntaxhighlight lang="bash"> ollama --version Warning: could not connect to a running Ollama instance Warning: client version is 0.35.0 </syntaxhighlight> === Intera..."

New page

== How to use ollama on the HPC ==
On the HPC we have <code>ollama</code> available through [[Environment Modules|modules]], in the GPU bucket.

To load the module, do this:<syntaxhighlight lang="bash">
ml GPU ollama
</syntaxhighlight>Once the module is loaded, <code>ollama</code> is available on your path:<syntaxhighlight lang="bash">
ollama --version
Warning: could not connect to a running Ollama instance
Warning: client version is 0.35.0
</syntaxhighlight>

=== Interactive session ===
To make working with ollama easier, we have created a small helper script, this will help you select a GPU, and set it up with the values you entered.<syntaxhighlight lang="bash">
menu_ollama.sh
</syntaxhighlight>Once you have entered all the values, the script will start an interactive job (so with output and input in your shell), which will look something like this:<syntaxhighlight lang="bash">
Now running:
srun --immediate=60 --nodes=1 --ntasks=1 --partition=gpu_amd --gres=gpu:1 --time=1:0:0 --mem-per-cpu=8G --cpus-per-gpu=8 /shared/apps/ollama/0.35.0/start_ollama.sh

srun: job 43006080 queued and waiting for resources
srun: job 43006080 has been allocated resources
remove ollama 0.35.0 binaries.
load ollama 0.35.0 binaries.
------------------------------------------------------------------------------
Use this on your host to setup a tunnel to the running instance:
ssh -L 11434:gpua201.internal.anunna.wur.nl:11435 haars001@login.anunna.wur.nl
Then connect to http://localhost:11434
For applications on other hosts within Anunna, you can use:
OLLAMA_HOST=gpua201.internal.anunna.wur.nl:11435
(Stop instance by pressing CTRL-C twice)
------------------------------------------------------------------------------
time=2026-10-02T10:32:43.672+02:00 level=INFO source=routes.go:2099 msg="server config" env="map[CUDA_VISIBLE_DEVICES:0 GGML_VK_VISIBLE_DEVICES: GPU_DEVICE_ORDINAL:0 HIP_VISIBLE_DEVICES: HSA_OVERRIDE_GFX_VERSION: HTTPS_PROXY: HTTP_PROXY: LLAMA_ARG_FIT: LLAMA_ARG_FIT_TARGET: NO_PROXY: OLLAMA_CONTEXT_LENGTH:0 OLLAMA_CREATE_REMOTE:false OLLAMA_DEBUG:INFO OLLAMA_DEBUG_LOG_REQUESTS:false OLLAMA_EDITOR: OLLAMA_FLASH_ATTENTION:false OLLAMA_GO_TEMPLATE:true OLLAMA_GPU_OVERHEAD:0 OLLAMA_HOST:http://0.0.0.0:11435 OLLAMA_IGPU_ENABLE: OLLAMA_KEEP_ALIVE:5m0s OLLAMA_KV_CACHE_TYPE: OLLAMA_LLM_LIBRARY: OLLAMA_LOAD_TIMEOUT:5m0s OLLAMA_MAX_LOADED_MODELS:0 OLLAMA_MAX_QUEUE:512 OLLAMA_MAX_TRANSFER_STREAMS:4 OLLAMA_MODELS:/lustre/scratch/GUESTS/haars001/ollama-models OLLAMA_NOHISTORY:false OLLAMA_NOPRUNE:false OLLAMA_NO_CLOUD:false OLLAMA_NUM_PARALLEL:1 OLLAMA_ORIGINS:[http://localhost https://localhost http://localhost:* https://localhost:* http://127.0.0.1 https://127.0.0.1 http://127.0.0.1:* https://127.0.0.1:* http://0.0.0.0 https://0.0.0.0 http://0.0.0.0:* https://0.0.0.0:* app://* file://* tauri://* vscode-webview://* vscode-file://*] OLLAMA_REMOTES:[ollama.com] OLLAMA_SCHED_SPREAD:false OLLAMA_VULKAN:true ROCR_VISIBLE_DEVICES:0 http_proxy: https_proxy: no_proxy:]"
time=2026-10-02T10:32:43.674+02:00 level=INFO source=routes.go:2101 msg="Ollama cloud disabled: false"
time=2026-10-02T10:32:44.055+02:00 level=INFO source=images.go:922 msg="total blobs: 0"
time=2026-10-02T10:32:44.055+02:00 level=INFO source=images.go:930 msg="total unused blobs removed: 0"
time=2026-10-02T10:32:44.063+02:00 level=INFO source=routes.go:2159 msg="Listening on 0.0.0.0:11435 (version 0.35.0)"
time=2026-10-02T10:32:44.074+02:00 level=INFO source=runner.go:60 msg="discovering available GPUs..."
time=2026-10-02T10:32:44.075+02:00 level=WARN source=runner.go:722 msg="user overrode visible devices" CUDA_VISIBLE_DEVICES=0
time=2026-10-02T10:32:44.075+02:00 level=WARN source=runner.go:722 msg="user overrode visible devices" ROCR_VISIBLE_DEVICES=0
time=2026-10-02T10:32:44.075+02:00 level=WARN source=runner.go:722 msg="user overrode visible devices" GPU_DEVICE_ORDINAL=0
time=2026-10-02T10:32:44.075+02:00 level=WARN source=runner.go:726 msg="if GPUs are not correctly discovered, unset and try again"
time=2026-10-02T10:32:44.325+02:00 level=INFO source=model_recommendations.go:177 msg="model recommendations cache sleep scheduled" wait=3h44m59.585365476s consecutive_failures=0
time=2026-10-02T10:32:49.707+02:00 level=INFO source=types.go:32 msg="inference compute" id=0 filter_id=0 library=ROCm compute=gfx90a name=ROCm0 description="AMD Instinct MI210" libdirs=ollama,rocm_v7_2 driver=0.0 pci_id=0000:67:00.0 type=discrete total="64.0 GiB" available="63.9 GiB"
time=2026-10-02T10:32:49.707+02:00 level=INFO source=routes.go:2209 msg="vram-based default context" total_vram="64.0 GiB" default_num_ctx=262144
</syntaxhighlight>As you can see, it will tell you how to connect to it, either from your computer, or within the cluster.

=== Batch wise usage ===
If you don't want or need to have <code>ollama</code> in an interactive session, you can bypass the menu, and just use the <code>start_ollama.sh</code> script directly:<syntaxhighlight lang="bash">
sbatch --ntasks=1 --partition=gpu_amd --gres=gpu:1 --time=1:0:0 --mem-per-cpu=8G --cpus-per-gpu=8 start_ollama.sh
</syntaxhighlight>Here we chose an AMD GPU, but one can of course also select Nvidia GPUs, that partition is called <code>gpu</code>.

After that you will have to look up the information on how to connect in the job output.

Workflows

Haars001 — Fri, 02 Oct 2026 08:24:19 GMT

← Older revision		Revision as of 08:24, 2 October 2026
Line 19:		Line 19:
	* Hybrid Workflows		* Hybrid Workflows
	* GPU workflows		* GPU workflows
			** [[ollama]]

JBrowse2

Honfi001 — Thu, 01 Oct 2026 08:57:45 GMT

← Older revision		Revision as of 08:57, 1 October 2026
(2 intermediate revisions by the same user not shown)
Line 1:		Line 1:
	{{DISPLAYTITLE:JBrowse 2}}		{{DISPLAYTITLE:JBrowse 2}}

	JBrowse 2 is a genome browser that runs in your web browser. On Anunna you can start it from ~~Open OnDemand~~, and it will show the data you already have on the cluster.		JBrowse 2 is a genome browser that runs in your web browser. On Anunna you can start it from the [https://apps.anunna.wur.nl Anunna Apps Portal], and it will show the data you already have on the cluster.

	This page explains what the JBrowse 2 app is, what each option on its form means, and which kinds of files it can show. It does not explain how to use JBrowse 2 itself. For that, see the [https://jbrowse.org/jb2/docs/ JBrowse 2 documentation].		This page explains what the JBrowse 2 app is, what each option on its form means, and which kinds of files it can show. It does not explain how to use JBrowse 2 itself. For that, see the [https://jbrowse.org/jb2/docs/ JBrowse 2 documentation].

JBrowse2

Honfi001 — Thu, 01 Oct 2026 08:46:32 GMT

Created page with "{{DISPLAYTITLE:JBrowse 2}} JBrowse 2 is a genome browser that runs in your web browser. On Anunna you can start it from Open OnDemand, and it will show the data you already have on the cluster. This page explains what the JBrowse 2 app is, what each option on its form means, and which kinds of files it can show. It does not explain how to use JBrowse 2 itself. For that, see the [https://jbrowse.org/jb2/docs/ JBrowse 2 documentation]. == Introduction == === What JBrow..."

New page

{{DISPLAYTITLE:JBrowse 2}}

JBrowse 2 is a genome browser that runs in your web browser. On Anunna you can start it from Open OnDemand, and it will show the data you already have on the cluster.

This page explains what the JBrowse 2 app is, what each option on its form means, and which kinds of files it can show. It does not explain how to use JBrowse 2 itself. For that, see the [https://jbrowse.org/jb2/docs/ JBrowse 2 documentation].

== Introduction ==

=== What JBrowse 2 is ===

A genome browser works a bit like a map application, but for a genome. The reference genome is the map. Your data is drawn on top of it, as '''tracks''': one track per file. You can zoom out to see a whole chromosome, or zoom in to see single bases.

JBrowse 2 can show, for example:

* sequencing reads aligned to the genome
* variants, such as SNPs
* coverage and other signals along the genome
* genes and other annotations

=== Why run it on Anunna ===

Sequencing data is large. A single alignment file can be many gigabytes, and you would not want to download it just to look at one gene.

With this app, you do not have to. JBrowse 2 runs on a compute node, right next to your data on [[Filesystems | Lustre]]. When you look at a region, only the small part of each file that covers that region is read. Nothing is copied.

Your files are also safe. The app can only '''read''' your data. It never changes, moves or adds anything in your folders.

== Starting the app ==

Go to the [https://apps.anunna.wur.nl Anunna Apps Portal], and find the '''JBrowse 2''' tile under ''All Applications''. Click on the button or selected it from the drop down, this will lead you to a form. Fill in the form, which is explained below, and click '''Launch'''.

Your session waits in the queue like any other job. Once it is running, a '''Connect to JBrowse 2''' button appears. Click it, and JBrowse 2 opens in a new browser tab with your data already loaded.

== The form ==

The form asks for two things about your data, the reference genome and a data folder, and a few things about the job itself.

{| class="wikitable"
! Option !! What it means !! Default
|-
| Maximum number of hours || How long the session may run. It stops after this time, even if you are still using it. Choose between 1 and 8 hours. || 4
|-
| Reference genome (indexed FASTA) || The genome your data is aligned to. '''Required.''' || —
|-
| Data folder to browse || The folder that holds your data files. ''Optional.'' || the folder of the reference
|-
| Number of CPU cores || How many cores the session gets. || 2
|-
| Total memory || How much memory the session gets. || 8 GB
|-
| Comment || A note for your job, for example a project number. || empty
|-
| SLURM options || Extra options for the job scheduler. Leave this empty if you are not sure. || empty
|-
| Email when started || Sends you an email when the session is ready. || off
|}

=== Reference genome ===

This is the most important choice. The reference is the "map" that everything else is drawn on, so '''all your data must be aligned to this same reference'''. If you choose a different version of the genome than the one your reads were aligned to, the tracks will not line up.

Use the browse button next to the field and choose a FASTA file. The file must:

* end in <code>.fa</code>, <code>.fasta</code> or <code>.fna</code>, or in <code>.fa.gz</code> (and so on) if it is compressed with <code>bgzip</code>
* have an index file next to it, with the same name plus <code>.fai</code>, for example <code>genome.fa</code> and <code>genome.fa.fai</code>
* if it is compressed, also have a <code>.gzi</code> file next to it

If the index is missing, the session stops right away and tells you so. See [[#Preparing your files|Preparing your files]] for how to make one.

=== Data folder ===

This is the folder with the files you want to see. Use the browse button next to the field and choose a folder, not a file.

The app looks through this folder '''and its subfolders''', up to four levels deep. Every file it can show becomes a track. Tracks are grouped by the subfolder they were found in, so a tidy folder gives you a tidy track list.

You can leave this field empty. The app then uses the folder that holds your reference genome. That is handy when the reference and the data are kept together.

A few tips:

* Choose the folder that holds the data you want to see, not a very large folder above it. The app adds at most 500 tracks. If a folder has more files than that, the rest are left out.
* Hidden files and folders, whose names start with a dot, are skipped.
* Subfolders that are only a link (a symlink) to somewhere else are not searched. Choose the real folder instead.

=== Cores and memory ===

A genome browser needs very little. JBrowse 2 does its drawing in your own web browser. The compute node only reads your files and sends the pieces you look at. The defaults, 2 cores and 8 GB, are enough for almost everyone. Asking for more will not make it faster, and it can make your session wait longer in the queue.

== Supported data ==

The app can show these kinds of files. Most of them need an '''index''' file next to them. The index is a small file that tells JBrowse 2 where each part of the genome is stored in the big file, so it can jump straight to the region you are looking at.

{| class="wikitable"
! Data !! File name ends in !! Index needed next to it !! Shown as
|-
| Aligned reads (BAM) || <code>.bam</code> || <code>.bai</code> or <code>.csi</code> || reads
|-
| Aligned reads (CRAM) || <code>.cram</code> || <code>.crai</code> || reads
|-
| Variants (VCF) || <code>.vcf.gz</code> || <code>.tbi</code> or <code>.csi</code> || variants
|-
| Annotation (GFF3) || <code>.gff3.gz</code> or <code>.gff.gz</code> || <code>.tbi</code> or <code>.csi</code> || features, such as genes
|-
| Regions (BED) || <code>.bed.gz</code> || <code>.tbi</code> or <code>.csi</code> || features
|-
| Signal (BigWig) || <code>.bw</code> or <code>.bigwig</code> || none || a graph, such as coverage
|-
| Regions (BigBed) || <code>.bb</code> or <code>.bigbed</code> || none || features
|}

Some things to know:

* '''Files without an index are skipped.''' The session still starts, but that file will not be in the track list.
* '''VCF, GFF3 and BED files must be compressed with <code>bgzip</code>.''' A plain <code>.vcf</code>, <code>.gff3</code> or <code>.bed</code> file is not shown. Note that <code>bgzip</code> is not the same as normal <code>gzip</code>: a file compressed with <code>gzip</code> cannot be indexed.
* '''CRAM files use the reference you chose''' to rebuild the reads. So for CRAM, choose the same reference genome the CRAM file was made with.
* Other file types in the folder, such as FASTQ or plain text, are simply ignored.

== Preparing your files ==

If your files are not indexed yet, you need to do this once. After that, the index files stay next to your data and you never need to do it again.

The tools you need are in the same module the app uses. Please do this in a job, not on the login node. A small job script is enough:

<syntaxhighlight lang="bash">
#!/bin/bash
#SBATCH --job-name=index-for-jbrowse
#SBATCH --time=1:00:00
#SBATCH --mem=4G
#SBATCH --cpus-per-task=1

module load 2025 JBrowse2/4.3.0-GCC-14.2.0

# Go to the folder that holds your files
cd /lustre/path/to/your/data

# The reference genome: makes genome.fa.fai
samtools faidx genome.fa

# Aligned reads: makes sample1.bam.bai
samtools index sample1.bam

# Variants: compress, then index. Makes variants.vcf.gz and variants.vcf.gz.tbi
bgzip variants.vcf
tabix -p vcf variants.vcf.gz
</syntaxhighlight>

Save this as, for example, <code>index.sh</code>, change the folder and file names to your own, and submit it with <code>sbatch index.sh</code>.

Use the line that matches each of your files:

{| class="wikitable"
! File !! Command
|-
| Reference genome || <code>samtools faidx genome.fa</code>
|-
| BAM or CRAM || <code>samtools index sample1.bam</code>
|-
| VCF || <code>bgzip variants.vcf</code>, then <code>tabix -p vcf variants.vcf.gz</code>
|-
| GFF3 || <code>bgzip genes.gff3</code>, then <code>tabix -p gff genes.gff3.gz</code>
|-
| BED || <code>bgzip regions.bed</code>, then <code>tabix -p bed regions.bed.gz</code>
|}

Indexing only works on files that are '''sorted by position''' along the genome. Alignment files made by most pipelines already are. If <code>samtools index</code> or <code>tabix</code> complains that the file is not sorted, sort it first, for example with <code>samtools sort</code> for a BAM file.

== If a file is missing from the track list ==

When the session starts, the app writes a short report in the session's <code>output.log</code>. You can open it from the session card in Open OnDemand, through the link to the session folder. The report lists every track that was added, and every file that was skipped, with the reason.

The usual reasons are:

* the index file is missing, or has a different name than the data file
* the file ends in <code>.vcf</code>, <code>.gff3</code> or <code>.bed</code> instead of <code>.vcf.gz</code> and so on
* the file is more than four folders below the data folder you chose
* the folder had more than 500 files the app could show

If the session does not start at all, the same log tells you why. Most often the reference genome has no <code>.fai</code> index, or the path you chose is not a FASTA file.

IntelMPI

Honfi001 — Fri, 18 Sep 2026 10:04:49 GMT

Running on a single node

← Older revision		Revision as of 10:04, 18 September 2026
Line 52:		Line 52:
	Nothing runs on the login nodes. Put the work in a script, keep data on Lustre under <code>$myScratch</code>, and submit with <code>sbatch</code>.		Nothing runs on the login nodes. Put the work in a script, keep data on Lustre under <code>$myScratch</code>, and submit with <code>sbatch</code>.

	<syntaxhighlight lang="bash">		<syntaxhighlight lang="bash">#!/bin/bash -l
	#!/bin/bash -l
	#SBATCH --job-name=impi_single		#SBATCH --job-name=impi_single
	#SBATCH --nodes=1		#SBATCH --nodes=1
Line 66:		Line 65:
	module load 2024 intel/2024a		module load 2024 intel/2024a

	mpirun -np ${SLURM_NTASKS} ./my_mpi_program ~~--input data.nc~~		export I_MPI_FABRICS=shm:ofi
	</syntaxhighlight>		export I_MPI_OFI_PROVIDER=psm2
			export FI_PROVIDER=psm2

			mpirun -np ${SLURM_NTASKS} ./my_mpi_program</syntaxhighlight>

	Two things carried over from [[OpenMPI]], for the same reasons: pass <code>${SLURM_NTASKS}</code> rather than hard-coding the rank count, and use <code>--mem-per-cpu</code> rather than <code>--mem</code>, which is a per-''node'' request divided among all the ranks on that node.		Two things carried over from [[OpenMPI]], for the same reasons: pass <code>${SLURM_NTASKS}</code> rather than hard-coding the rank count, and use <code>--mem-per-cpu</code> rather than <code>--mem</code>, which is a per-''node'' request divided among all the ranks on that node.
Line 94:		Line 96:

	mpirun -genv I_MPI_OFI_PROVIDER psm2 -genv FI_PROVIDER psm2 \		mpirun -genv I_MPI_OFI_PROVIDER psm2 -genv FI_PROVIDER psm2 \
	-np ${SLURM_NTASKS} ./my_mpi_program ~~--input data.nc~~		-genv I_MPI_FABRICS=shm:ofi -np ${SLURM_NTASKS} ./my_mpi_program
	</syntaxhighlight>		</syntaxhighlight>