School-lab GPUs disappear from nvidia-smi after a driver change. What should ops do before reimaging?
Select an answer to reveal the explanation.
Short Explanation
If school-lab GPUs vanish, nvidia-smi is the attendance sheet—confirm driver visibility before the reimage steamroller. Ordering replacements on a hunch skips evidence.
Full Explanation
Using nvidia-smi evidence to confirm driver/GPU visibility issues on civic nodes is the right step when school-lab GPUs disappear from nvidia-smi after a driver change—before reimaging. Missing devices in nvidia-smi commonly indicate driver/module mismatch, failed install, persistence issues, or hardware not enumerated. Confirming that evidence scopes repair to driver rollback, reinstall, or a single bad host instead of mass lab destruction.
Reimaging every lab overnight without confirming GPU visibility evidence fails by change-control: reimage is a heavy hammer that erases forensic clues. Assuming GPUs were stolen and ordering replacements immediately fails by premature hardware replacement without software visibility checks after a driver change. Disabling nvidia-smi so missing GPUs cannot be documented fails by ticket integrity—you need the tool that proves absence or presence.
Exam caveat: nvidia-smi confirmation is necessary before reimage, not a guarantee that reimage will never be needed. If drivers are correct and devices still missing, then hardware, PCIe, or power issues rise in priority—with evidence preserved.
Operational check: on affected lab nodes, capture nvidia-smi, dmesg/driver version, and install logs in the change ticket; compare to a known-good node; remediate the driver visibility fault; only escalate to reimage or RMA after evidence rules in those paths.