reanimator/bee
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

webui/topo: read from techdump instead of live nvidia-smi, fix socket/NUMA mismatch, fix cross-tier SAT status merge

Browse Source 
/topo blocked the HTTP request on live nvidia-smi calls with no timeout
(topo -m, nvlink -s/-e run on every page load), so a wedged driver hung
the page indefinitely. CaptureTechnicalDump now persists these dumps
once per audit cycle; the page reads them from techdump/ instead.

buildSocketIndex mapped NUMA node number to CPU by treating dmidecode's
Socket Designation (often 1-indexed, "CPU1"/"CPU2") as equal to the
NUMA node number (always 0-indexed) — GPUs/NICs on NUMA node 0 fell
into the "unknown" column, others attached to the wrong CPU box. Now
ranks CPUs by Socket value instead of assuming a shared numbering base.

Fixed a bug in ComponentStatusDB/applyComponentStatusDB where GPU SAT
results were keyed per-target ("pcie:gpu:nvidia-stress") instead of
per-vendor, which both broke cross-tier severity tracking (a later
clean "2. Check" run and an earlier failing "3. Load" run never
compared severities) and silently failed to match any real BDF, so the
DB overlay never reached the topology graph at all. GPU keys are now
normalized to vendor ("pcie:gpu:nvidia"/"pcie:gpu:amd"). Also skip
writing to the DB when a SAT task was aborted by the user (ctx
canceled), so a partial run can't stomp a previously recorded status.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
Mikhail Chusavitin
2026-07-07 17:14:03 +03:00
co-authored by Claude Sonnet 5
parent f46fc98110
commit 2d84ddb577
8 changed files with 186 additions and 55 deletions
Download Patch File Download Diff File Expand all files Collapse all files
audit/internal/app/component_status_db.go
+13 -4
View File
@@ -177,12 +177,21 @@ func ApplySATResultToDB(db *ComponentStatusDB, target, archivePath string) {
source := "sat:" + target
dbStatus := satStatusToDBStatus(overall)
// Map SAT target to component keys.
// Map SAT target to component keys. GPU targets are keyed by vendor, not
// by the raw target string: "nvidia" (Check tier) and "nvidia-stress" /
// "nvidia-targeted-stress" (Load/Burn tier) all exercise the same
// physical GPUs, so they must share one severity-tracked record — a
// severity-1 Check run after a severity-3 Load failure must not lose
// that failure just because it ran more recently. Recording each target
// under its own key (the previous behavior) also silently broke
// applyComponentStatusDB below, which expects "pcie:gpu:<vendor>" and
// otherwise fails to match any real BDF.
switch target {
case "nvidia", "nvidia-targeted-stress", "nvidia-compute", "nvidia-targeted-power", "nvidia-pulse",
"nvidia-interconnect", "nvidia-bandwidth", "amd", "nvidia-stress",
"amd-stress", "amd-mem", "amd-bandwidth":
db.Record("pcie:gpu:"+target, source, dbStatus, target+" SAT: "+overall)
"nvidia-interconnect", "nvidia-bandwidth", "nvidia-stress":
db.Record("pcie:gpu:nvidia", source, dbStatus, target+" SAT: "+overall)
case "amd", "amd-stress", "amd-mem", "amd-bandwidth":
db.Record("pcie:gpu:amd", source, dbStatus, target+" SAT: "+overall)
case "memory", "memory-stress", "sat-stress":
db.Record("memory:all", source, dbStatus, target+" SAT: "+overall)
case "cpu", "platform-stress":