Run directly on live CD: bee-smoketest Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
185 lines
5.5 KiB
Bash
Executable File
185 lines
5.5 KiB
Bash
Executable File
#!/bin/sh
|
|
# smoketest.sh — run on a live ISO via SSH to verify all critical components.
|
|
#
|
|
# Usage:
|
|
# ssh root@<ip> 'sh -s' < smoketest.sh
|
|
# or: scp smoketest.sh root@<ip>:/var/tmp/ && ssh root@<ip> sh /var/tmp/smoketest.sh
|
|
#
|
|
# Exit code: 0 = all required checks passed, 1 = at least one required check failed.
|
|
|
|
PASS=0
|
|
FAIL=0
|
|
WARN=0
|
|
|
|
ok() { echo "[ OK ] $*"; PASS=$((PASS+1)); }
|
|
fail() { echo "[ FAIL ] $*"; FAIL=$((FAIL+1)); }
|
|
warn() { echo "[ WARN ] $*"; WARN=$((WARN+1)); }
|
|
info() { echo "[ INFO ] $*"; }
|
|
|
|
echo "========================================"
|
|
echo " bee live ISO smoketest"
|
|
echo " host: $(uname -n)"
|
|
echo " kernel: $(uname -r)"
|
|
echo " date: $(date -u)"
|
|
echo "========================================"
|
|
echo ""
|
|
|
|
# --- kernel version ---
|
|
KVER=$(uname -r)
|
|
info "kernel: $KVER"
|
|
|
|
# --- PATH ---
|
|
echo "-- PATH & binaries --"
|
|
for tool in dmidecode smartctl nvme ipmitool lspci audit; do
|
|
if p=$(PATH="/usr/local/bin:$PATH" command -v "$tool" 2>/dev/null); then
|
|
ok "$tool found: $p"
|
|
else
|
|
fail "$tool: NOT FOUND"
|
|
fi
|
|
done
|
|
|
|
if p=$(PATH="/usr/local/bin:$PATH" command -v nvidia-smi 2>/dev/null); then
|
|
ok "nvidia-smi found: $p"
|
|
else
|
|
fail "nvidia-smi: NOT FOUND"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- NVIDIA modules --"
|
|
KO_DIR="/usr/local/lib/nvidia"
|
|
if [ -d "$KO_DIR" ]; then
|
|
ko_count=$(ls "$KO_DIR"/*.ko 2>/dev/null | wc -l)
|
|
ok "NVIDIA ko dir exists: $KO_DIR ($ko_count .ko files)"
|
|
else
|
|
fail "NVIDIA ko dir missing: $KO_DIR"
|
|
fi
|
|
|
|
for mod in nvidia nvidia_modeset nvidia_uvm; do
|
|
if /sbin/lsmod 2>/dev/null | grep -q "^$mod "; then
|
|
ok "module loaded: $mod"
|
|
else
|
|
fail "module NOT loaded: $mod"
|
|
fi
|
|
done
|
|
|
|
echo ""
|
|
echo "-- nvidia-smi --"
|
|
if PATH="/usr/local/bin:$PATH" command -v nvidia-smi >/dev/null 2>&1; then
|
|
if PATH="/usr/local/bin:$PATH" nvidia-smi -L 2>/dev/null | grep -q "GPU"; then
|
|
gpu_count=$(PATH="/usr/local/bin:$PATH" nvidia-smi -L 2>/dev/null | grep -c "GPU")
|
|
ok "nvidia-smi: $gpu_count GPU(s) found"
|
|
else
|
|
fail "nvidia-smi: runs but no GPUs detected"
|
|
fi
|
|
else
|
|
fail "nvidia-smi: not found in PATH"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- lib symlinks --"
|
|
for lib in libnvidia-ml libcuda; do
|
|
if [ -f "/usr/lib/${lib}.so.1" ] || [ -L "/usr/lib/${lib}.so.1" ]; then
|
|
ok "/usr/lib/${lib}.so.1 exists"
|
|
else
|
|
fail "/usr/lib/${lib}.so.1 MISSING (nvidia-smi will fail)"
|
|
fi
|
|
done
|
|
|
|
echo ""
|
|
echo "-- gcompat (glibc compat for nvidia-smi) --"
|
|
if [ -L /lib64/ld-linux-x86-64.so.2 ] || [ -f /lib64/ld-linux-x86-64.so.2 ]; then
|
|
ok "gcompat: /lib64/ld-linux-x86-64.so.2 present"
|
|
else
|
|
fail "gcompat: /lib64/ld-linux-x86-64.so.2 MISSING — nvidia-smi will fail to exec"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- openrc services --"
|
|
for svc in bee-nvidia bee-network; do
|
|
if rc-service "$svc" status >/dev/null 2>&1; then
|
|
ok "service running: $svc"
|
|
else
|
|
fail "service NOT running: $svc"
|
|
fi
|
|
done
|
|
|
|
for svc in bee-audit-debug dropbear bee-sshsetup; do
|
|
if [ -f "/etc/init.d/$svc" ]; then
|
|
if rc-service "$svc" status >/dev/null 2>&1; then
|
|
ok "service running: $svc"
|
|
else
|
|
warn "service not running: $svc (may be one-shot)"
|
|
fi
|
|
fi
|
|
done
|
|
|
|
echo ""
|
|
echo "-- audit binary --"
|
|
AUDIT=/usr/local/bin/audit
|
|
if [ -x "$AUDIT" ]; then
|
|
ok "audit binary: present"
|
|
ver=$("$AUDIT" --version 2>/dev/null || "$AUDIT" version 2>/dev/null || echo "unknown")
|
|
info "audit version: $ver"
|
|
else
|
|
fail "audit binary: NOT FOUND at $AUDIT"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- audit last run --"
|
|
# audit binary logs via slog to stderr (bee-audit.log); JSON output goes to bee-audit.json.
|
|
# slog format: time=... level=INFO msg="audit output written" path=...
|
|
if [ -f /var/log/bee-audit.json ] && [ -s /var/log/bee-audit.json ]; then
|
|
ok "audit: bee-audit.json present and non-empty"
|
|
info "size: $(du -sh /var/log/bee-audit.json | cut -f1)"
|
|
else
|
|
fail "audit: bee-audit.json missing or empty"
|
|
fi
|
|
|
|
if [ -f /var/log/bee-audit.log ]; then
|
|
last_line=$(tail -1 /var/log/bee-audit.log)
|
|
info "last log line: $last_line"
|
|
# slog writes: msg="audit output written" on success
|
|
if grep -q "audit output written" /var/log/bee-audit.log 2>/dev/null; then
|
|
ok "audit: completed successfully"
|
|
else
|
|
warn "audit: 'audit output written' not found in log — may have failed"
|
|
fi
|
|
# check for nvidia enrichment skip (slog message from nvidia collector)
|
|
if grep -q "nvidia: enrichment skipped\|nvidia.*skipped\|enrichment skipped" /var/log/bee-audit.log 2>/dev/null; then
|
|
reason=$(grep -E "nvidia.*skipped|enrichment skipped" /var/log/bee-audit.log | tail -1)
|
|
fail "audit: nvidia enrichment skipped — $reason"
|
|
else
|
|
ok "audit: nvidia enrichment OK (no skip message)"
|
|
fi
|
|
else
|
|
warn "audit: no log found at /var/log/bee-audit.log"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- network --"
|
|
if ip route show default 2>/dev/null | grep -q "default"; then
|
|
gw=$(ip route show default | awk '{print $3}' | head -1)
|
|
ok "default route: $gw"
|
|
else
|
|
fail "no default route"
|
|
fi
|
|
if ping -c1 -W3 1.1.1.1 >/dev/null 2>&1; then
|
|
ok "internet: reachable (1.1.1.1)"
|
|
else
|
|
fail "internet: unreachable"
|
|
fi
|
|
|
|
echo ""
|
|
echo "-- /etc/profile.d/bee.sh PATH --"
|
|
if grep -q "/usr/local/bin" /etc/profile.d/bee.sh 2>/dev/null; then
|
|
ok "/etc/profile.d/bee.sh exports /usr/local/bin"
|
|
else
|
|
fail "/etc/profile.d/bee.sh does not add /usr/local/bin to PATH"
|
|
fi
|
|
|
|
echo ""
|
|
echo "========================================"
|
|
echo " Results: OK=$PASS FAIL=$FAIL WARN=$WARN"
|
|
echo "========================================"
|
|
[ "$FAIL" -eq 0 ] && exit 0 || exit 1
|