This website requires JavaScript.
196acf0b9f
feat(iso): auto-queue an Audit task once bee-web is up
Mikhail Chusavitin and Claude Sonnet 5
2026-09-15 18:55:48 +03:00
b09e94d02a
fix(sat): serialize recurring IPMI polling behind one shared telemetry pipeline
Mikhail Chusavitin and Claude Sonnet 5
2026-09-15 18:55:43 +03:00
cfef94870f
test: cover IPMI, PSU, and SAT formatting
mchus
2026-09-12 16:36:07 +03:00
9da6dac1bd
refactor: remove dead updater and simplify ISO launcher
mchus
2026-09-12 16:29:48 +03:00
9c5c29239c
fix(iso): let empty directories survive the squashfs layer split
mchus and Claude Sonnet 5
2026-09-12 16:19:36 +03:00
78d1b9b599
fix(iso): seed a real file in /var/log/nvidia-dcgm so it survives squashfs layering
mchus and Claude Sonnet 5
2026-09-12 16:11:24 +03:00
bbc6fb1c36
fix(nvidia): own /var/log/nvidia-dcgm so dcgmi diag deployment check passes
mchus and Claude Sonnet 5
2026-09-12 11:53:52 +03:00
5190d50358
fix: tolerate unavailable CPU sensor telemetry
mchus
2026-09-08 09:49:40 +03:00
b8c45d54c1
feat(iso): split the live medium into semantic SquashFS layers
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 15:38:14 +03:00
bc57b85d3f
Merge feat/fan-ceiling-check: fan ceiling check + topology fan/PSU tiles
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 14:34:22 +03:00
08044c81e7
fix(webui): don't show PSU load % until the observed peak clears idle draw
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 12:56:25 +03:00
7ed652c5b1
feat: autotune PSU capacity from the same full-load run as fan ceilings
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 12:53:13 +03:00
e713504fc9
feat(webui): show PSU load as a fill on the topology PSU card
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 12:37:59 +03:00
59271fa674
feat(webui): break the topology PSU row into one card per supply
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 12:32:56 +03:00
a4853a0a4f
feat(webui): live-update topology fan tiles without a page reload
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 12:09:37 +03:00
d8b05069e0
fix(webui): move Fan Ceiling Check to the CPU/Memory/Storage row
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 11:43:41 +03:00
8cb250f3f4
fix(sat): fan check uses hottest GPU load + IPMI-hang-proof polling
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 11:02:31 +03:00
bb2a501a28
feat(sat): fan ceiling check + topology fan tiles
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 10:35:34 +03:00
e4f7519ef3
feat(webui): show GPU serial in NVIDIA selection pickers
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 10:04:47 +03:00
f31971f440
fix(webui): make metrics history immune to system-clock changes
Mikhail Chusavitin and Claude Sonnet 5
2026-09-04 10:04:13 +03:00
4953bb33b2
docs: set 16 GB minimum supported RAM
Mikhail Chusavitin
2026-09-04 09:14:38 +03:00
677f4f4006
fix(webui): keep Check-only tasks out of Load
Mikhail Chusavitin
2026-09-03 15:41:18 +03:00
a9a9b18a8c
fix(webui): format server and browser time consistently
Mikhail Chusavitin
2026-09-03 15:41:11 +03:00
b14417b729
fix(runtime): avoid Copy to RAM worker panic
Mikhail Chusavitin
2026-09-03 15:34:30 +03:00
eae5570730
fix(iso): stop fast-path repack from clobbering merged-usr /lib
Mikhail Chusavitin and Claude Sonnet 5
2026-09-03 15:29:48 +03:00
489dd2830c
fix(iso): retry toram RAM copy with exponential backoff
Mikhail Chusavitin and Claude Sonnet 5
2026-09-03 10:11:05 +03:00
d0ed1bda91
fix(iso): hash build scripts into the fast-path safety check
Mikhail Chusavitin and Claude Sonnet 5
2026-09-03 15:32:33 +03:00
76db45ee95
feat(audit): capture IPMI FRU/Elabel, mc info and LAN on every audit
Mikhail Chusavitin and Claude Sonnet 5
2026-09-03 15:18:01 +03:00
319f08ac4b
feat(topo): per-socket bar layout with disks under their controller
Mikhail Chusavitin and Claude Sonnet 5
2026-09-03 12:15:49 +03:00
b1ab866f58
fix(runtime): make Copy to RAM switch live backing
Mikhail Chusavitin
2026-09-03 10:47:15 +03:00
347bc8310a
fix(pcie): verify GPU links under real bandwidth load
Mikhail Chusavitin
2026-09-03 10:22:42 +03:00
642e68631d
fix(iso): render isolinux flavour before validation
Mikhail Chusavitin
2026-09-01 15:22:31 +03:00
0a6ca8ba0f
refactor: harden diagnostics and consolidate runtime code
Mikhail Chusavitin
2026-09-01 13:01:28 +03:00
ac4bc0b2b7
refactor: modularize audit and harden build validation
Mikhail Chusavitin
2026-08-31 21:22:16 +03:00
bb22ccfafe
docs(bundle): document pcie-link and tpm SAT targets in bundle README
Mikhail Chusavitin and Claude Sonnet 5
2026-08-28 11:34:45 +03:00
366c1d5f29
fix(pcie): compare retrain verdict against peer-capped max speed
Mikhail Chusavitin and Claude Sonnet 5
2026-08-28 11:30:57 +03:00
c5c7a2dea3
fix(iso): serialize coldplug and harden DMA isolation
Mikhail Chusavitin
2026-08-28 09:51:51 +03:00
2354599889
fix(pcie): account for downstream link capability
Mikhail Chusavitin
2026-08-28 09:12:02 +03:00
af95216c6f
feat(tpm): add read-only TPM validation
Mikhail Chusavitin
2026-08-26 17:13:50 +03:00
9e466b8a70
fix(pcie): stop misreporting empty/vanished slots as speed degradation
Mikhail Chusavitin and Claude Sonnet 5
2026-08-26 12:44:45 +03:00
ccc781856e
fix: complete hardware collection diagnostics
mchus
2026-08-25 17:04:33 +03:00
33ace33de5
chore: bump bible and internal/chart submodule pointers
mchus and Claude Sonnet 5
2026-08-25 16:25:53 +03:00
d2ac42dd7f
build(iso): switch default Debian mirror, add --mirror override
mchus and Claude Sonnet 5
2026-08-25 16:22:26 +03:00
ba250330d5
fix(app): capture PCIe link / AER diagnostics at boot, not just on demand
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 18:39:28 +03:00
b11018ac5e
fix(collector): stop pinning GPU PCIe status on an unverified idle reading
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 18:39:19 +03:00
7aa276320b
fix(webui): trust IPMI FRU readback over ipmitool's exit code
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 18:39:02 +03:00
198567dffe
fix: surface CPU thermal throttling in status and SAT results
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 17:41:40 +03:00
56d12b1f3c
fix(collector): use word-boundary matching for dmesg severity
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 17:41:28 +03:00
a09201c7ab
fix(collector): recover storage model for NVMe drives exposed as SCSI
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 17:41:23 +03:00
9add5616dd
fix(app): bound support-bundle subprocess calls with timeouts
Mikhail Chusavitin and Claude Sonnet 5
2026-08-24 17:41:18 +03:00
134bed3eae
refactor: dedupe status-severity ranking and small webui helpers
Mikhail Chusavitin and Claude Sonnet 5
2026-08-18 14:44:52 +03:00
e3697c0a11
fix(pcie): stop misclassifying same-vendor non-GPU devices as GPUs
Mikhail Chusavitin and Claude Sonnet 5
2026-08-17 12:50:50 +03:00
4ac9863353
fix(webui): keep ramp-up burns as a single multi-GPU task
Mikhail Chusavitin and Claude Sonnet 5
2026-08-17 10:59:49 +03:00
b1f165edb3
fix(collector): surface PCIe link-speed degradation in component-status DB, sample it under load
Mikhail Chusavitin and Claude Sonnet 5
2026-08-06 13:51:14 +03:00
3ec7ca08da
feat(webui): sync host clock and timezone from browser
Mikhail Chusavitin and Claude Sonnet 5
2026-08-06 13:43:31 +03:00
b2b3f86c8d
fix(nvidia): surface Xid 79/154 GPU bus-fall-off as a physical-reboot-required signal
Mikhail Chusavitin and Claude Sonnet 5
2026-08-06 12:28:34 +03:00
a34e823f82
fix(collector): stop NVIDIA enrichment from clobbering PCIe status
Mikhail Chusavitin and Claude Sonnet 5
2026-08-04 17:28:47 +03:00
6c8be629d2
fix(nvidia): load ib_umad and install nvlsm/libnvidia-nscq for NVL5 fabric
Mikhail Chusavitin and Claude Sonnet 5
2026-07-30 10:34:11 +03:00
983f41a1f0
fix(nvidia): detect NVSwitch fabric via driver, not host-PCIe
mchus
2026-07-29 18:58:19 +03:00
8a91f0f783
fix(webui): repair broken scenario Run button onclick, dedupe build.sh overlay staging
mchus and Claude Sonnet 5
2026-07-29 18:11:24 +03:00
1045fa9118
build(iso): drop gsp-off from boot menu, fail-safe keeps GPUs functional
mchus and Claude Opus 4.8
2026-07-29 14:57:35 +03:00
9b5d4e3036
build(iso): redefine no-GUI and fail-safe boot entries
mchus and Claude Opus 4.8
2026-07-29 14:23:25 +03:00
95d3532bb9
build(iso): unify GRUB/isolinux boot menus, add iommu=pt to fail-safe
mchus and Claude Opus 4.8
2026-07-29 14:11:43 +03:00
a9924b08b0
feat(scenario): add nvbandwidth-acs-ab, an ACS off/on A/B for cross-socket P2P
mchus and Claude Opus 4.8
2026-07-29 13:45:34 +03:00
3ed826aeb7
fix(cli): route bee run scenarios through App so blackbox mirrors them
mchus and Claude Opus 4.8
2026-07-29 13:45:34 +03:00
86aa11c2d7
fix(webui): dispatch "scenario" tasks in the real task runner, collapse duplicate target switch
mchus and Claude Opus 4.8
2026-07-29 13:00:41 +03:00
d108df7fe9
platform/webui: add scenario description, show it in the Scenario page's list
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 19:20:31 +03:00
5a780e96b4
platform/app/webui: ship the power-watch scenario baked into the image
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 19:10:35 +03:00
bcf02e0515
webui: add "6. Scenario" page — run scriptable test scenarios from removable media through the normal task queue
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 18:58:11 +03:00
20cd317c87
platform/cmd/webui: add a scriptable test-scenario engine, load scenarios from blackbox USB, GPU status detail inventory fallback
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 17:51:45 +03:00
49979c4da4
app: don't let a renamed-away source path abort the whole blackbox/support-bundle copy
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 16:28:11 +03:00
fa618d5abb
platform/app: restructure support bundle/blackbox mirror into device-type export/, status/, tasks/, livecd/
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 14:33:26 +03:00
e03267a72f
platform/app: fix nvbandwidth split fallback, blackbox discovery churn, and add blocking sync brackets around load steps
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 14:12:54 +03:00
6d13c17d36
iso: give the live image a real hostname instead of live-build's default "debian"
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 10:30:40 +03:00
e036a9c7b8
boot: drop pci=realloc from default/toram GRUB entries, fold into troubleshooting entry
Mikhail Chusavitin and Claude Sonnet 5
2026-07-28 09:34:25 +03:00
781cf5dcbf
platform/app: stream SAT job output live and kick blackbox on job completion
Mikhail Chusavitin
2026-07-27 17:40:22 +03:00
ced2175fb0
platform: split nvbandwidth SAT into per-socket + all-GPU passes
Mikhail Chusavitin
2026-07-27 17:01:39 +03:00
41f683de2b
webui/topo: attach memory DIMMs to their CPU column in the topology diagram
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 17:15:06 +03:00
c1f0f7824e
platform: mirror per-drive disk reports into techdump
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 16:38:07 +03:00
b30d34199a
app/webui: replace generic "see summary.txt" SAT failure text with the real reason
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 15:48:12 +03:00
5aee146903
app: ship a README.md in every support bundle and blackbox capture
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 11:37:44 +03:00
cfaa15ec7c
platform/webui: fold confidential-computing into a GPU config + NVLink check
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 11:36:52 +03:00
cc3997f7b1
app/webui: fix component-status.json multi-process clobber and unbounded growth
Mikhail Chusavitin and Claude Sonnet 5
2026-07-09 11:24:18 +03:00
1d5c02ebaa
webui/topo: responsive PSU/firmware layout, stacked group cards, fix zero-CPU crash
Mikhail Chusavitin
2026-07-09 10:52:38 +03:00
10557ec0f6
webui/collector: strip ANSI escapes in nvidia-smi topo parsing, auto-prepare drives for RAID mirror creation
mchus and Claude Fable 5
2026-07-08 23:16:04 +03:00
1175e6ccd8
gpu: collect reset-required/row-remap status and classify Xid codes by severity
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 19:28:59 +03:00
d850d4fc3a
iso: update vendor storcli2 to 08.017.000.0008
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 18:05:21 +03:00
d2dd305c54
iso: harden fast-path staleness check and add DCGM version verification
Mikhail Chusavitin
2026-07-07 18:03:31 +03:00
4e306ff78c
raid: parse storcli64/storcli2's actual per-slot drive JSON shape
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 17:21:05 +03:00
2d84ddb577
webui/topo: read from techdump instead of live nvidia-smi, fix socket/NUMA mismatch, fix cross-tier SAT status merge
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 17:14:03 +03:00
f46fc98110
sat: add tests for informational discovery jobs and retry logic
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 15:31:31 +03:00
53c46465d2
webui: add persistent GPU settings management (ECC/MIG/CC/power limit)
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 15:25:25 +03:00
2599d9c5e3
sat/nvidia-load: make dcgmi discovery non-fatal and bound systemctl restarts
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 15:25:14 +03:00
b7f015c713
raid: add storcli2 support for Tri-Mode controllers (SAS3808-iMR/9500 series)
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 15:25:06 +03:00
a3377083aa
webui: add /topo server topology page
Mikhail Chusavitin and Claude Sonnet 5
2026-07-07 11:50:33 +03:00
33d6eee9cf
sat: collect NVLink status/errors, fix nv-hostengine restart via systemd
Mikhail Chusavitin
2026-07-07 10:21:56 +03:00
a98721b038
sat: collect dcgmi discovery + full nvvs debug log for all DCGM diags
Mikhail Chusavitin and Claude Sonnet 5
2026-07-06 16:52:55 +03:00
10b45bc267
iso: explicitly pin datacenter-gpu-manager-4-core
Mikhail Chusavitin and Claude Sonnet 5
2026-07-06 14:15:35 +03:00
fb581c28d8
iso: fix driver pin — 580.126.20 404s on the installer CDN
Mikhail Chusavitin and Claude Sonnet 5
2026-07-06 13:58:15 +03:00
ffb85d55a2
iso: pin NVIDIA driver/DCGM to DGX OS 7.5.0-validated combo
Mikhail Chusavitin and Claude Sonnet 5
2026-07-06 13:33:52 +03:00
mchus