⚠️ pandoc “file not found in resource path” when rendering Rmd on HPC (e.g., generating HTML reports with eyeris::bidsify())

[shawntz]

Summary

This note is for users rendering R Markdown on HPC environments where pandoc is centrally installed (e.g., via shared software modules) and encounter errors like the one below when generating HTML reports using eyeris::bidsify():

/share/software/user/open/pandoc/2.7.3/bin/pandoc +RTS -K512m -RTS sub-02.knit.md --to html4 --from markdown+autolink_bare_uris+tex_math_single_backslash --output sub-02.html --lua-filter /share/software/user/open/R/4.2.0/lib64/R/library/rmarkdown/rmarkdown/lua/pagebreak.lua --lua-filter /share/software/user/open/R/4.2.0/lib64/R/library/rmarkdown/rmarkdown/lua/latex-div.lua --self-contained --variable bs3=TRUE --standalone --section-divs --table-of-contents --toc-depth 3 --variable toc_float=1 --variable toc_selectors=h1,h2,h3 --variable toc_collapsed=1 --variable toc_smooth_scroll=1 --variable toc_print=1 --template /share/software/user/open/R/4.2.0/lib64/R/library/rmarkdown/rmd/h/default.html --no-highlight --variable highlightjs=1 --variable theme=bootstrap --css /home/users/shawnsch/R/x86_64-pc-linux-gnu-library/4.2/eyeris/rmarkdown/css/report.css --mathjax --variable 'mathjax-url=https://mathjax.rstudio.com/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML' --include-in-header /tmp/RtmpRSI8rj/rmarkdown-str1d32793b83a6.html 

This issue is especially likely when using --self-contained HTML rendering with rmarkdown::render(), which is the method eyeris::bidsify() uses to generate its HTML reports.

Cause

In many HPC setups, pandoc is an older version (e.g., v2.7.3), and files like CSS, JS, or figures referenced from within a package or user directory are not accessible or get dropped during the rendering pipeline.

In my own testing, I’ve frequently encountered this issue when:

• Using the --self-contained flag (as set by default in rmarkdown::render())
• Rendering within a SLURM job or other headless batch context with strict permissions

For example, on Stanford’s Sherlock HPC, users are encouraged to load software via shared modules, yet the only available pandoc version is v2.7.3, which is quite outdated 😭.

Solution / Workaround

1. Install a local copy of the latest version of pandoc, which at the time of this writing is v3.1.11.1:

# create a local installation directory
cd $HOME
mkdir -p local/pandoc
cd local/pandoc

# download the tar.gz for Linux x86_64 (static binary)
curl -L -o pandoc.tar.gz https://github.com/jgm/pandoc/releases/download/3.1.11.1/pandoc-3.1.11.1-linux-amd64.tar.gz

# extract
tar -xzf pandoc.tar.gz --strip-components=1

# add pandoc to path
export PATH="$HOME/local/pandoc/bin:$PATH"

# and then test it with
which pandoc
pandoc --version

2. In your headless Rscript workflow, call the following function before running any eyeris commands. This ensures the R Markdown render engine uses your locally installed version of pandoc:

rmarkdown::find_pandoc(cache = FALSE, dir = "~/local/pandoc/bin", version = "3.1.11.1")

And voilà, the eyeris HTML reports should now generate on your headless HPC setup! Hope this helps anyone else running into similar issues with pandoc rendering in constrained environments.

If you encounter related problems or discover cleaner workarounds, feel free to comment here so others can benefit. 🧠💡

This issue is being left open as a searchable note but is not something that requires a fix within the package itself.