Classify exit-1 errors & guard telemetry

Analyze logs for generic exit code 1 and export an ERROR_CATEGORY_OVERRIDE so telemetry receives a more accurate error category (apt, oom, network, storage, dependency). Preserve any existing TELEMETRY_TYPE when posting updates. Add defense-in-depth by disabling strict error traps before running grep/sed log analysis to avoid spurious error_handler invocations. Mark successful installs with INSTALL_COMPLETE and update the error handler to only report a successful "done" telemetry state when INSTALL_COMPLETE is explicitly set, preventing false-positive success reports from early zero-exit exits.

misc/api.func

@@ -348,10 +348,10 @@ explain_exit_code() {
 json_escape() {
   # Escape a string for safe JSON embedding using awk (handles any input size).
   # Pipeline: strip ANSI → remove control chars → escape \ " TAB → join lines with \n
-  printf '%s' "$1" \
+  printf '%s' "$1" |
-    | sed 's/\x1b\[[0-9;]*[a-zA-Z]//g' \
+    sed 's/\x1b\[[0-9;]*[a-zA-Z]//g' |
-    | tr -d '\000-\010\013\014\016-\037\177\r' \
+    tr -d '\000-\010\013\014\016-\037\177\r' |
-    | awk '
+    awk '
         BEGIN { ORS = "" }
         {
           gsub(/\\/, "\\\\")   # backslash → \\
@@ -627,8 +627,8 @@ post_to_api() {
 
   [[ "${DEV_MODE:-}" == "true" ]] && echo "[DEBUG] post_to_api() DIAGNOSTICS=$DIAGNOSTICS RANDOM_UUID=$RANDOM_UUID NSAPP=$NSAPP" >&2
 
-  # Set type for later status updates
+  # Set type for later status updates (preserve if already set, e.g. turnkey)
-  TELEMETRY_TYPE="lxc"
+  TELEMETRY_TYPE="${TELEMETRY_TYPE:-lxc}"
 
   local pve_version=""
   if command -v pveversion &>/dev/null; then
@@ -692,6 +692,7 @@ EOF
   # Send initial "installing" record with retry.
   # This record MUST exist for all subsequent updates to succeed.
   local http_code="" attempt
+  local _post_success=false
   for attempt in 1 2 3; do
     if [[ "${DEV_MODE:-}" == "true" ]]; then
       http_code=$(curl -sS -w "%{http_code}" -m "${TELEMETRY_TIMEOUT}" -X POST "${TELEMETRY_URL}" \
@@ -703,11 +704,19 @@ EOF
         -H "Content-Type: application/json" \
         -d "$JSON_PAYLOAD" -o /dev/null 2>/dev/null) || http_code="000"
     fi
-    [[ "$http_code" =~ ^2[0-9]{2}$ ]] && break
+    if [[ "$http_code" =~ ^2[0-9]{2}$ ]]; then
+      _post_success=true
+      break
+    fi
     [[ "$attempt" -lt 3 ]] && sleep 1
   done
 
-  POST_TO_API_DONE=true
+  # Only mark done if at least one attempt succeeded.
+  # If all 3 failed, POST_TO_API_DONE stays false so post_update_to_api
+  # and on_exit() know the initial record was never created.
+  # The server has fallback logic to create a new record on status updates,
+  # so subsequent calls can still succeed even without the initial record.
+  POST_TO_API_DONE=${_post_success}
 }
 
 # ------------------------------------------------------------------------------
@@ -798,15 +807,19 @@ EOF
 
   # Send initial "installing" record with retry (must succeed for updates to work)
   local http_code="" attempt
+  local _post_success=false
   for attempt in 1 2 3; do
     http_code=$(curl -sS -w "%{http_code}" -m "${TELEMETRY_TIMEOUT}" -X POST "${TELEMETRY_URL}" \
       -H "Content-Type: application/json" \
       -d "$JSON_PAYLOAD" -o /dev/null 2>/dev/null) || http_code="000"
-    [[ "$http_code" =~ ^2[0-9]{2}$ ]] && break
+    if [[ "$http_code" =~ ^2[0-9]{2}$ ]]; then
+      _post_success=true
+      break
+    fi
     [[ "$attempt" -lt 3 ]] && sleep 1
   done
 
-  POST_TO_API_DONE=true
+  POST_TO_API_DONE=${_post_success}
 }
 
 # ------------------------------------------------------------------------------
@@ -1083,6 +1096,12 @@ EOF
 # - Used to group errors in dashboard
 # ------------------------------------------------------------------------------
 categorize_error() {
+  # Allow build.func to override category based on log analysis (exit code 1 subclassification)
+  if [[ -n "${ERROR_CATEGORY_OVERRIDE:-}" ]]; then
+    echo "$ERROR_CATEGORY_OVERRIDE"
+    return
+  fi
+
   local code="$1"
   case "$code" in
   # Network errors (curl/wget)

misc/build.func

@@ -222,9 +222,12 @@ update_motd_ip() {
     local current_ip="$(hostname -I | awk '{print $1}')"
 
     # Escape sed special chars in replacement strings (& \ |)
-    current_os="${current_os//\\/\\\\}"; current_os="${current_os//&/\\&}"
+    current_os="${current_os//\\/\\\\}"
-    current_hostname="${current_hostname//\\/\\\\}"; current_hostname="${current_hostname//&/\\&}"
+    current_os="${current_os//&/\\&}"
-    current_ip="${current_ip//\\/\\\\}"; current_ip="${current_ip//&/\\&}"
+    current_hostname="${current_hostname//\\/\\\\}"
+    current_hostname="${current_hostname//&/\\&}"
+    current_ip="${current_ip//\\/\\\\}"
+    current_ip="${current_ip//&/\\&}"
 
     # Update only if values actually changed
     if ! grep -q "OS:.*$current_os" "$PROFILE_FILE" 2>/dev/null; then
@@ -4223,6 +4226,53 @@ EOF'
       fi
     fi
 
+    # Defense-in-depth: Ensure error handling stays disabled during recovery.
+    # Some functions (e.g. silent/$STD) unconditionally re-enable set -Eeuo pipefail
+    # and trap 'error_handler' ERR. If any code path above called such a function,
+    # the grep/sed pipelines below would trigger error_handler on non-match (exit 1).
+    set +Eeuo pipefail
+    trap - ERR
+
+    # --- Exit code 1 subclassification: analyze logs BEFORE telemetry call ---
+    # Exit code 1 is generic ("General error"). Analyze logs to determine the
+    # real error category so telemetry gets a useful classification instead of "shell".
+    local is_oom=false
+    local is_network_issue=false
+    local is_apt_issue=false
+    local is_cmd_not_found=false
+    local is_disk_full=false
+
+    if [[ $install_exit_code -eq 1 && -f "$combined_log" ]]; then
+      if grep -qiE 'E: Unable to|E: Package|E: Failed to fetch|dpkg.*error|broken packages|unmet dependencies|dpkg --configure -a' "$combined_log"; then
+        is_apt_issue=true
+      fi
+      if grep -qiE 'Cannot allocate memory|Out of memory|oom-killer|Killed process|JavaScript heap' "$combined_log"; then
+        is_oom=true
+      fi
+      if grep -qiE 'Could not resolve|DNS|Connection refused|Network is unreachable|No route to host|Temporary failure resolving|Failed to fetch' "$combined_log"; then
+        is_network_issue=true
+      fi
+      if grep -qiE ': command not found|No such file or directory.*/s?bin/' "$combined_log"; then
+        is_cmd_not_found=true
+      fi
+      if grep -qiE 'ENOSPC|no space left on device|Disk quota exceeded|errno -28' "$combined_log"; then
+        is_disk_full=true
+      fi
+    fi
+
+    # Set override for categorize_error() so telemetry gets the real category
+    if [[ "$is_apt_issue" == true ]]; then
+      export ERROR_CATEGORY_OVERRIDE="dependency"
+    elif [[ "$is_oom" == true ]]; then
+      export ERROR_CATEGORY_OVERRIDE="resource"
+    elif [[ "$is_network_issue" == true ]]; then
+      export ERROR_CATEGORY_OVERRIDE="network"
+    elif [[ "$is_disk_full" == true ]]; then
+      export ERROR_CATEGORY_OVERRIDE="storage"
+    elif [[ "$is_cmd_not_found" == true ]]; then
+      export ERROR_CATEGORY_OVERRIDE="dependency"
+    fi
+
     # Report failure to telemetry API (now with log available on host)
     # NOTE: Do NOT use msg_info/spinner here — the background spinner process
     # causes SIGTSTP in non-interactive shells (bash -c "$(curl ...)"), which
@@ -4231,13 +4281,6 @@ EOF'
     post_update_to_api "failed" "$install_exit_code"
     $STD echo -e "${TAB}${CM:-✔} Failure reported"
 
-    # Defense-in-depth: Ensure error handling stays disabled during recovery.
-    # Some functions (e.g. silent/$STD) unconditionally re-enable set -Eeuo pipefail
-    # and trap 'error_handler' ERR. If any code path above called such a function,
-    # the grep/sed pipelines below would trigger error_handler on non-match (exit 1).
-    set +Eeuo pipefail
-    trap - ERR
-
     # Show combined log location
     if [[ -n "$CTID" && -n "${SESSION_ID:-}" ]]; then
       msg_custom "📋" "${YW}" "Installation log: ${combined_log}"
@@ -4266,12 +4309,9 @@ EOF'
     # Prompt user for cleanup with 60s timeout
     echo ""
 
-    # Detect error type for smart recovery options
+    # Extend error detection for non-exit-1 codes (exit 1 was already analyzed above)
-    local is_oom=false
+    # The is_* flags were set above for exit code 1 log analysis; here we add
-    local is_network_issue=false
+    # exit-code-specific detections for other codes.
-    local is_apt_issue=false
-    local is_cmd_not_found=false
-    local is_disk_full=false
     local error_explanation=""
     if declare -f explain_exit_code >/dev/null 2>&1; then
       error_explanation="$(explain_exit_code "$install_exit_code")"
@@ -4321,26 +4361,6 @@ EOF'
       ;;
     esac
 
-    # Exit 1 subclassification: analyze logs to identify actual root cause
-    # Many exit 1 errors are actually APT, OOM, network, or command-not-found issues
-    if [[ $install_exit_code -eq 1 && -f "$combined_log" ]]; then
-      if grep -qiE 'E: Unable to|E: Package|E: Failed to fetch|dpkg.*error|broken packages|unmet dependencies|dpkg --configure -a' "$combined_log"; then
-        is_apt_issue=true
-      fi
-      if grep -qiE 'Cannot allocate memory|Out of memory|oom-killer|Killed process|JavaScript heap' "$combined_log"; then
-        is_oom=true
-      fi
-      if grep -qiE 'Could not resolve|DNS|Connection refused|Network is unreachable|No route to host|Temporary failure resolving|Failed to fetch' "$combined_log"; then
-        is_network_issue=true
-      fi
-      if grep -qiE ': command not found|No such file or directory.*/s?bin/' "$combined_log"; then
-        is_cmd_not_found=true
-      fi
-      if grep -qiE 'ENOSPC|no space left on device|Disk quota exceeded|errno -28' "$combined_log"; then
-        is_disk_full=true
-      fi
-    fi
-
     # Show error explanation if available
     if [[ -n "$error_explanation" ]]; then
       echo -e "${TAB}${RD}Error: ${error_explanation}${CL}"
@@ -4542,6 +4562,7 @@ EOF'
 
           if [[ $apt_retry_code -eq 0 ]]; then
             msg_ok "Installation completed successfully after APT repair!"
+            INSTALL_COMPLETE=true
             post_update_to_api "done" "0" "force"
             return 0
           else
@@ -5716,6 +5737,7 @@ EOF
     systemctl start ping-instances.service
   fi
 
+  INSTALL_COMPLETE=true
   post_update_to_api "done" "none"
 }
 

misc/error_handler.func

@@ -507,14 +507,23 @@ _stop_container_if_installing() {
 on_exit() {
   local exit_code=$?
 
-  # Report orphaned "installing" records to telemetry API
+  # Report orphaned telemetry records
-  # Catches ALL exit paths: errors, signals, AND clean exits where
+  # Two scenarios handled:
-  # post_to_api was called but post_update_to_api was never called
+  # 1. POST_TO_API_DONE=true but POST_UPDATE_DONE=false: Record was created but
-  if [[ "${POST_TO_API_DONE:-}" == "true" && "${POST_UPDATE_DONE:-}" != "true" ]]; then
+  #    never got a final status update → send abort/done now.
-    if [[ $exit_code -ne 0 ]]; then
+  # 2. POST_TO_API_DONE=false but DIAGNOSTICS=yes: Initial post failed (server
-      _send_abort_telemetry "$exit_code"
+  #    unreachable/timeout), but the server has fallback create-on-update logic,
-    elif declare -f post_update_to_api >/dev/null 2>&1; then
+  #    so a status update can still create the record. Worth one last try.
-      post_update_to_api "done" "0" 2>/dev/null || true
+  if [[ "${POST_UPDATE_DONE:-}" != "true" ]]; then
+    if [[ "${POST_TO_API_DONE:-}" == "true" || "${DIAGNOSTICS:-no}" == "yes" ]]; then
+      if [[ $exit_code -ne 0 ]]; then
+        _send_abort_telemetry "$exit_code"
+      elif [[ "${INSTALL_COMPLETE:-}" == "true" ]] && declare -f post_update_to_api >/dev/null 2>&1; then
+        # Only report success if the install was explicitly marked complete.
+        # Without this guard, early bailouts (e.g. user cancelled) with exit 0
+        # would be falsely reported as successful installations.
+        post_update_to_api "done" "0" 2>/dev/null || true
+      fi
     fi
   fi