tracel-ai/burn · error

Could not get cache directory

Error message

Could not get cache directory

What it means

This is the `expect` on `dirs::cache_dir()` in the DISTS weights module's `get_cache_dir()`, used by `load_pretrained_weights` to locate `~/.cache/burn-dataset/dists`. It panics when the OS-level user cache directory cannot be determined (no `$XDG_CACHE_HOME`/`$HOME` on Linux, or no platform cache path). Any DISTS pretrained-weight load then aborts before downloading the VGG16 backbone.

Source

Thrown at crates/burn-train/src/metric/vision/dists/weights.rs:22

use burn_store::{ModuleSnapshot, PytorchStore};
use std::fs::{File, create_dir_all};
use std::io::Write;
use std::path::PathBuf;

use super::metric::Dists;

/// URL for pretrained DISTS alpha/beta weights from the official repository.
/// Reference: https://github.com/dingkeyan93/DISTS
const DISTS_WEIGHTS_URL: &str =
    "https://github.com/dingkeyan93/DISTS/raw/master/DISTS_pytorch/weights.pt";

/// URL for ImageNet pretrained VGG16 backbone weights from PyTorch.
const VGG16_IMAGENET_URL: &str = "https://download.pytorch.org/models/vgg16-397923af.pth";

/// Get the cache directory for DISTS weights.
fn get_cache_dir() -> PathBuf {
    let cache_dir = dirs::cache_dir()
        .expect("Could not get cache directory")
        .join("burn-dataset")
        .join("dists");

    if !cache_dir.exists() {
        create_dir_all(&cache_dir).expect("Failed to create cache directory");
    }

    cache_dir
}

/// Download file if not cached.
fn download_if_needed(url: &str, cache_path: &PathBuf, message: &str) {
    if !cache_path.exists() {
        let bytes = download_file_as_bytes(url, message);
        let mut file = File::create(cache_path).expect("Failed to create cache file");
        file.write_all(&bytes).expect("Failed to write weights");
    }
}

View on GitHub (pinned to d16f7ba2ed)

Solutions

1. Export `XDG_CACHE_HOME=/some/writable/dir` (or set HOME) before launching the program
2. In Docker, add `ENV XDG_CACHE_HOME=/cache` plus a created/mounted directory
3. Pre-provision the dists cache so the environment issue is caught early, or run with a fixed env via a wrapper script
4. Patch to fall back to `std::env::temp_dir()` when `dirs::cache_dir()` returns None

Example fix

// before (systemd unit)
[Service]
ExecStart=/usr/local/bin/trainer
// after (systemd unit)
[Service]
Environment=XDG_CACHE_HOME=/var/cache/trainer
ExecStart=/usr/local/bin/trainer

fn ensure_dists_cache_base() -> std::path::PathBuf {
    let base = dirs::cache_dir()
        .or_else(|| std::env::var("XDG_CACHE_HOME").ok().map(std::path::PathBuf::from))
        .unwrap_or_else(std::env::temp_dir);
    std::fs::create_dir_all(&base).expect("cache base unusable");
    base
}
if dirs::cache_dir().is_none() { eprintln!("set XDG_CACHE_HOME before loading DISTS weights"); }

fn cache_dir_available() -> bool { dirs::cache_dir().is_some() }

let weights = std::panic::catch_unwind(load_pretrained_weights)
    .map_err(|_| anyhow!("DISTS weight load panicked; ensure XDG_CACHE_HOME/HOME is set"))?;

When it happens

Trigger: Calling `load_pretrained_weights` for the DISTS metric on a host where `dirs::cache_dir()` is None — Linux process with HOME and XDG_CACHE_HOME unset (minimal Docker image, systemd unit, CI job, cron).

Common situations: Containers running as root with scrubbed env; daemons/services without a user session; stripped-down CI images; platforms where the `dirs` crate has no cache-dir convention.

Related errors

• Could not get cache directory
• Could not get cache directory
• Failed to create cache directory
• Failed to create cache directory
• Failed to create cache directory

AI-assisted analysis of tracel-ai/burn@d16f7ba2ed (2026-09-05). Data as JSON: /api/errors/b008e05e30ca435b. Report an issue: GitHub.