 Alexander Korolev
|
e00a59ec49
Log probe SSH failures (previously swallowed as Err(_))
|
преди 1 седмица |
|
Alexander Korolev
|
b19f8aea3b
health/web: cluster-level semantics — one dead node or switch off = cluster OFF
|
преди 2 месеца |
|
Alexander Korolev
|
8446509baf
Global config: recipe_sources + cluster_sources
|
преди 2 месеца |
|
Alexander Korolev
|
551ea7d8d7
feat(nccl): auto-detect NCCL_IB_GID_INDEX per node, opt-out via config pin
|
преди 2 месеца |
|
Alexander Korolev
|
0181645b9e
fix: make IB config optional for single-node clusters
|
преди 2 месеца |
|
Alexander Korolev
|
a3bd00eef8
fix: tmux serve quoting, venv setup, cluster probe on dgx3.lan
|
преди 2 месеца |
|
Alexander Korolev
|
1b4d07309a
fix: make NCCL config optional for single-node, capture stderr, support --on for tmux
|
преди 2 месеца |
|
Alexander Korolev
|
82bbb916d2
fix(i18n): set thread-local locale in spawned tasks
|
преди 2 месеца |
|
Alexander Korolev
|
1ba1d99e05
fix: resolve review findings across modules
|
преди 2 месеца |
|
Alexander Korolev
|
2b0deeaffa
feat(recipe): extra_mounts with config alias resolution + tilde + env injection
|
преди 3 месеца |
|
Alexander Korolev
|
b37592678f
feat: drop_caches (echo 3 | sudo tee /proc/sys/vm/drop_caches) before start
|
преди 3 месеца |
|
Alexander Korolev
|
31aaf141e5
refactor: split ends (HA/SSH) from control logic with traits and orchestrator
|
преди 3 месеца |
|
Alexander Korolev
|
7158e63649
refactor: separate cluster state from SSH probing for testability
|
преди 3 месеца |
