Part of epic tracebloc/backend#1142 / cli#355. Implemented by PR #359.
Bare tb resources shows the allocation in two lines + a pointer to the scriptable set — no inline prompt, no physical-host line:
Your secure environment is equipped with: 8 CPU · 32 GiB
A training run is allocated up to: 4 CPU · 16 GiB
Do you want to change the allocation? Run `tb resources set` (guided on a terminal).
- Line 1 — secure environment total: what the cluster can schedule.
- Line 2 — per-run ceiling: the per-training allocation (cluster-wide, uniform — the only CLI-settable dial).
- Change it:
tb resources set (flags for scripting; guided walkthrough on a TTY). Bare resources never mutates. --verbose: per-node breakdown + raw values.
History / scope change (2026-07-21)
Originally scoped as a 3-layer view (machine / secure environment / per-run) plus an inline guided-change wizard. Per Lukas's "just two lines" direction this was simplified to the two-line view above. Dropped:
- the "Your machine has:" physical-host line (the "this machine" vs "tracebloc uses" framing confused users), and
- the inline
[y/N] prompt (replaced by a plain pointer to tb resources set).
The full 3-layer spec and its implementation live in the closed PR #361 if we ever revisit the richer view.
Part of epic tracebloc/backend#1142 / cli#355. Implemented by PR #359.
Bare
tb resourcesshows the allocation in two lines + a pointer to the scriptableset— no inline prompt, no physical-host line:tb resources set(flags for scripting; guided walkthrough on a TTY). Bareresourcesnever mutates.--verbose: per-node breakdown + raw values.History / scope change (2026-07-21)
Originally scoped as a 3-layer view (machine / secure environment / per-run) plus an inline guided-change wizard. Per Lukas's "just two lines" direction this was simplified to the two-line view above. Dropped:
[y/N]prompt (replaced by a plain pointer totb resources set).The full 3-layer spec and its implementation live in the closed PR #361 if we ever revisit the richer view.