Fetch — pull files from remote clusters

The fetch command copies files from a remote cluster to your local machine. It reads every [[clusters.fetch_dirs]] block in the config and, for each pair, mirrors the remote directory into its local destination, skipping files that are already up-to-date.

Tip

Use the sbatchman remotes-config TUI to manage you configurations.

---

CLI Usage

# Pull all clusters, all fetch_dirs
sbatchman fetch

# Pull only one cluster
sbatchman fetch -c cluster-a

# Dry-run to see what would be transferred (rsync only)
sbatchman fetch --dry-run

# Use SFTP instead of rsync (e.g. no rsync on the login node)
sbatchman fetch -b sftp

---

How it works

1. sbatchman loads ~/.config/sbatchman/remotes-config.toml.
2. For each matching cluster it reads the [[clusters.fetch_dirs]] blocks.
3. Each block describes one remote → local pair.
4. Files are copied only when the remote version is newer than the local one (--update / mtime comparison), so repeated runs are fast.
5. Progress is streamed to the terminal in real time.

Note

The fetch feature and configuration is really similar to sync. Please refer to the Sync page for more details.

---

Config reference

[global]
transfer_backend = "rsync"   # default backend for all clusters

# Excluded from BOTH fetch and sync
common_excludes = [".git", ".venv", "__pycache__", "build"]

# Excluded from fetch only
fetch_excludes  = ["results", "data", "datasets", "plots"]

[[clusters]]
name     = "cluster-a"
host     = "login.cluster-a.com"
port     = 22
user     = "alice"
key_path = "~/.ssh/id_ed25519"

# Excluded from all operations on this cluster
excludes       = ["scratch"]

# Excluded from fetch only on this cluster
fetch_excludes = ["logs"]

[[clusters.fetch_dirs]]
aliad    = "myproject"            # short name used with -a / --aliases
remote   = "~/myproject"          # path on the cluster
local    = "~/results/myproject"  # where to put it locally
excludes = ["tmp"]                # excluded for this pair only

Tip: remote paths starting with ~/ are resolved on the remote shell, so they respect the user's actual home directory even on non-standard systems.

---

Examples

# Pull two specific clusters
sbatchman fetch -c cluster-a -c cluster-b -a myapp

# Dry-run on a single cluster with sftp
sbatchman fetch -c cluster-a -b sftp --dry-run