sparkrun logs

sparkrun logs <target> [options]

Show the logs of a running workload, with docker logs / journalctl semantics. TARGET can be a recipe name or a cluster ID (from sparkrun status output).

Called bare, it dumps the workload’s logs and exits. Use -f to stay attached and stream new output — Ctrl+C detaches and never kills the inference job.

Options

Option	Description
--hosts / -H	Comma-separated host list
--hosts-file	File with hosts (one per line, # comments)
--cluster	Use a saved cluster by name
--tp / --tensor-parallel	Match host trimming from run
--port	Override port (to match run-time override)
--served-model-name	Override served model name (to match run-time override)
-n / --lines / --tail	Number of log lines to show (default: all). With -f, sets the scrollback shown before following.
-f / --follow	Stay attached and stream new lines
-a / --all-sources	Read every worker/rank, not just the head

Reading workers

By default sparkrun logs reads the head. -a / --all-sources adds every worker:

• Without -f, each source is dumped in full — head first, then workers by rank.
• With -f, the streams interleave in arrival order.

# Dump everything the head has logged
sparkrun logs qwen3-1.7b-vllm --cluster mylab

# Last 200 lines, then follow
sparkrun logs qwen3-1.7b-vllm --cluster mylab -n 200 -f

# Follow head and all workers together
sparkrun logs e5f6a7b8 -f -a

Changed in 0.3.0

--tail previously defaulted to 100 lines and always followed. It is now -n / --lines / --tail, defaults to showing all lines, and following is opt-in via -f.

Tip

Some runtimes launch the server with docker exec inside a long-lived container, so the container’s own stdout is not the workload’s log. sparkrun resolves the right log location per runtime, which is why sparkrun logs finds output that a plain docker logs on the host would miss.