Fix NVIDIA self-heal recovery flow

This commit is contained in:
Mikhail Chusavitin
2026-04-20 09:43:22 +03:00
parent 1cfabc9230
commit 84a2551dc0
4 changed files with 215 additions and 4 deletions

View File

@@ -0,0 +1,30 @@
package platform
import (
"fmt"
"os/exec"
"time"
)
const nvidiaRecoverHelper = "/usr/local/bin/bee-nvidia-recover"
func runNvidiaRecover(args ...string) (string, error) {
helperArgs := append([]string{nvidiaRecoverHelper}, args...)
if _, err := exec.LookPath("systemd-run"); err == nil {
unit := fmt.Sprintf("bee-nvidia-recover-%d", time.Now().UnixNano())
cmdArgs := []string{
"systemd-run",
"--quiet",
"--pipe",
"--wait",
"--collect",
"--service-type=oneshot",
"--unit", unit,
}
cmdArgs = append(cmdArgs, helperArgs...)
raw, err := exec.Command("sudo", cmdArgs...).CombinedOutput()
return string(raw), err
}
raw, err := exec.Command("sudo", helperArgs...).CombinedOutput()
return string(raw), err
}