icinga2-checks/check_scrutiny_disks.py

#!/usr/bin/env python3
import argparse
import subprocess
import sys
from datetime import datetime, timedelta
from typing import List

import requests

from checker import nagios
from checker.http import get_with_retry


def get_disk_wwn_ids(ignore_non_smart: bool = False) -> List[str] or bool:
    wwn_ids = []
    try:
        output = subprocess.check_output(["sudo", "smartctl", "--scan"])
        for line in output.decode("utf-8").strip().split("\n"):
            parts = line.split()
            if len(parts) >= 3:
                device = parts[0]
                device_type = parts[2].replace('scsi', 'sat,auto')
                try:
                    smart_supported = subprocess.check_output(
                        ["sudo", "smartctl", "-i", device, "-d", device_type]).decode("utf-8")
                    if "SMART support is: Enabled" in smart_supported:
                        wwn_line = [line for line in smart_supported.split("\n") if "LU WWN Device Id" in line]
                        wwn_id = '0x' + wwn_line[0].replace('LU WWN Device Id: ', '').replace(' ', '')
                        wwn_ids.append(wwn_id)
                    # else:
                    #     # TODO: warn if a drive doesn't support SMART
                except subprocess.CalledProcessError as e:
                    if ignore_non_smart:
                        continue
                    else:
                        print(f"UNKNOWN: subprocess Error - {e}")
                        sys.exit(nagios.UNKNOWN)
    except subprocess.CalledProcessError as e:
        print(f"UNKNOWN: subprocess Error - {e}")
        sys.exit(nagios.UNKNOWN)
    return wwn_ids


def get_smart_health(wwn_id: str, scrutiny_endpoint: str) -> dict:
    url = f"{scrutiny_endpoint}/api/device/{wwn_id}/details"
    response = get_with_retry(url)
    if response.status_code == 200:
        return response.json()
    elif response.status_code == 404:
        return {
            'fetch_error': f"Disk {wwn_id} not found on Scrutiny"
        }
    else:
        return {
            'fetch_error': f"Error {response.status_code} for disk {wwn_id}: {response.text}"
        }


def main(args):
    results = {}
    wwn_ids = get_disk_wwn_ids(args.ignore_non_smart)
    metrics_out_of_date = False
    for wwn_id in wwn_ids:
        smart_health = get_smart_health(wwn_id, args.scrutiny_endpoint)

        if smart_health.get('fetch_error'):
            print('UNKNOWN -', smart_health.get('fetch_error'))
            sys.exit(nagios.UNKNOWN)

        name = f'/dev/{smart_health["data"]["device"]["device_name"]} {wwn_id}'  # differentiate disks in RAID arrays

        results[name] = {
            'wwn_id': wwn_id,
            'failed_attributes': [],
        }

        metadata = smart_health['metadata']

        # For testing
        # smart_health['data']['device']['UpdatedAt'] = '2023-04-28T23:00:03.071184465Z'

        last_updated = datetime.strptime(smart_health['data']['device']['UpdatedAt'][:-4] + 'Z', '%Y-%m-%dT%H:%M:%S.%fZ')
        if datetime.utcnow() - timedelta(hours=args.time_delta_limit) > last_updated:
            metrics_out_of_date = True

        if smart_health and len(smart_health['data']['smart_results']):
            try:
                disk_data = smart_health['data']['smart_results'][0]['attrs']
            except Exception as e:
                print('UNKNOWN - failed to parse data:', e)
                print('Key "data":', smart_health['data'].keys())
                print('Key "smart_results":', len(smart_health['data']['smart_results']))
                print('Key "attrs":', smart_health['data']['smart_results'][0].keys())
                sys.exit(nagios.UNKNOWN)
            for attribute_id, values in disk_data.items():
                if values['status'] == 0:
                    continue
                # elif values['status'] == 2 and not args.warn_non_critical:
                #     continue
                values['attribute_name'] = metadata[attribute_id]['display_name']
                values['metadata'] = metadata[attribute_id]
                if 'observed_thresholds' in values['metadata'].keys():
                    del values['metadata']['observed_thresholds']
                results[name]['failed_attributes'].append(values)
            results[name]['status'] = 'good'
        else:
            results[name]['status'] = 'no data'

    crit_disks = {}
    warn_disks = {}
    for disk, values in results.items():
        if values['status'] != 'no data':
            for item in values['failed_attributes']:
                if item['status'] == 2 and args.warn_non_critical:
                    if disk not in warn_disks.keys():
                        warn_disks[disk] = []
                    warn_disks[disk].append({
                        'raw_value': item['raw_value'],
                        'display_name': item['metadata']['display_name']
                    })
                if item['status'] == 4:
                    if disk not in crit_disks.keys():
                        crit_disks[disk] = []
                    crit_disks[disk].append({
                        'raw_value': item['raw_value'],
                        'display_name': item['metadata']['display_name']
                    })
        else:
            if disk not in warn_disks.keys():
                warn_disks[disk] = []
            warn_disks[disk].append({
                'raw_value': 'no data',
                'display_name': 'no data'
            })

    dt = '<dt>' if args.html else ''
    dts = '</dt>' if args.html else ''
    dd = '<dd>' if args.html else '\t- '
    dds = '</dd>' if args.html else ''

    out_of_date_str = f'metrics are >{args.time_delta_limit} hrs out of date ' if metrics_out_of_date else ''

    return_code = nagios.OK
    if len(crit_disks):
        return_code = nagios.CRITICAL
        if len(warn_disks):
            x = f' and {len(warn_disks)} {"warnings" if len(results) > 1 else "warning"}'
        else:
            x = ''
        print(
            f'CRITICAL: {out_of_date_str + "and " if len(out_of_date_str) else ""}{len(crit_disks)} {"errors" if len(crit_disks) > 1 else "error"}{x}')

        print('<dl>')
        print(f'{dt}Disks with Errors:{dts}')

        for disk, warns in crit_disks.items():
            if args.html:
                disk_name = f'- <a href="{args.pretty_url}/web/device/{results[disk]["wwn_id"]}" target="_blank">{disk}</a>'
            else:
                disk_name = f'\t- {disk}'
            print(f'{dd}{disk_name}: {", ".join([x["display_name"] for x in warns])}{dds}')

        print('</dl>', end='')

    if len(out_of_date_str) and not len(crit_disks):
        return_code = nagios.CRITICAL
        w = "warnings" if len(warn_disks) > 1 else "warning"
        print(f'CRITICAL: {out_of_date_str}{"and " + str(len(warn_disks)) + " " + w if len(warn_disks) else ""}')

    if len(warn_disks):
        if return_code < nagios.CRITICAL:
            return_code = nagios.WARNING
            print(f'WARNING: {len(warn_disks)} {"warnings" if len(warn_disks) > 1 else "warning"}')

        print('<dl>')
        print(f'{dt}Disks with issues:{dts}')
        for disk, warns in warn_disks.items():
            if args.html:
                disk_name = f'- <a href="{args.pretty_url}/web/device/{results[disk]["wwn_id"]}" target="_blank">{disk}</a>'
            else:
                disk_name = f'\t- {disk}'
            print(f'{dd}{disk_name}: {", ".join([x["display_name"] for x in warns])}{dds}')
        print('</dl>', end='')

    if not len(crit_disks) and not len(warn_disks):
        print(f'OK: all {len(results.keys())} {"disks" if len(results.keys()) > 1 else "disk"} are healthy!', end='')

    print(f"|'warnings'={len(warn_disks)};;; 'errors'={len(crit_disks)};;; 'num_disks'={len(results.keys())};;;")
    sys.exit(return_code)


def is_smartmontools_installed():
    try:
        result = subprocess.run(['dpkg', '-s', 'smartmontools'], stdout=subprocess.PIPE, stderr=subprocess.PIPE,
                                check=True, text=True)
        if 'Status: install ok installed' in result.stdout:
            return True
        else:
            return False
    except subprocess.CalledProcessError:
        return False


if __name__ == "__main__":
    parser = argparse.ArgumentParser(description='')
    parser.add_argument('--scrutiny-endpoint', required=True, help='Base URL for scrutiny.')
    parser.add_argument('--time-delta-limit', default=24, type=int,
                        help='The Scrutiny data must not be older than this many hours. Default: 24.')
    parser.add_argument('--warn-non-critical', action='store_true',
                        help='Warn when a non-critical metric is marked as failed.')
    parser.add_argument('--html', action='store_true', help='Print HTML.')
    parser.add_argument('--pretty-url', help='The pretty URL to link to when printing HTML.')
    parser.add_argument('--ignore-non-smart', action='store_true',
                        help="Ignore any non-SMART devices and any devices that error when reading SMART.")
    parser.add_argument('--dont-warn-no-data', action='store_true', help="Don't warn if there is no data for a disk.")
    args = parser.parse_args()

    if args.html and not args.pretty_url:
        print('UKNOWN - when using --html you must also set --pretty-url')
        sys.exit(nagios.UNKNOWN)

    if not is_smartmontools_installed():
        print('UNKNOWN - smartmontools is not installed.')
        sys.exit(nagios.UNKNOWN)

    args.scrutiny_endpoint = args.scrutiny_endpoint.strip('/')
    args.pretty_url = args.pretty_url.strip('/') if args.pretty_url else None
    try:
        main(args)
    except Exception as e:
        print(f'UNKNOWN: exception "{e}"')
        import traceback

        print(traceback.format_exc())
        sys.exit(nagios.UNKNOWN)