File: //lib/check_mk_agent/plugins/mk_sap.py
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# Copyright (C) 2019 Checkmk GmbH - License: GNU General Public License v2
# This file is part of Checkmk (https://checkmk.com). It is subject to the terms and
# conditions defined in the file COPYING, which is part of this source code package.
__version__ = "2.3.0p38"
# This agent plugin has been built to collect information from SAP R/3 systems
# using RFC calls. It needs the python module pyrfc.
# pyrfc requires following libs (installable via pip3: Sphinx Cython setuptools-git
# via apt or yum/dnf: gcc gcc-c++ python3-devel) (and the nwrfcsdk (can be downloaded from SAP
# download portal) installed to be working. To install pyrfc export the SAPNWRFC_HOME:
# "export SAPNWRFC_HOME=/usr/sap/nwrfcsdk". You can configure the agent plugin
# using the configuration file /etc/check_mk/sap.cfg (a sample file can be
# found in Checkmk git at agents/sap/sap.cfg) to tell it how to connect to
# your SAP instance and which values you want to fetch from your system to be
# forwarded to and checked by Checkmk.
#
# Tested Versions:
# pyrfc:
# 2.1.0, 2.4.0
# SAP nwrfc:
# 750P_8-70002752
# required params in sap.cfg:
# 'ashost''sysnr' 'client' 'user' 'passwd' 'trace' 'loglevel' 'lang'
#
# create file /etc/ld.so.conf.d/pyrfc.conf and containing the path
# "/usr/sap/nwrfcsdk/lib" and run "ldconfig" afterwards.
#
# During development the "CCMS_Doku.pdf" was really helpful.
import ast
import datetime
import fcntl
import fnmatch
import os
import sys
import time
try:
from typing import Any # noqa: F401 # pylint: disable=unused-import
except ImportError:
pass
if sys.version_info[:2] < (3, 5):
RecursionError = RuntimeError # pylint: disable=redefined-builtin
# #############################################################################
# This sign is used to separate the path parts given in the config
SEPARATOR = "/"
# This are the different classes of monitoring objects which
# can be found in the tree.
#
# Summarizs information from several subnodes
MTE_SUMMARY = "050"
# A monitoring object which has several subnodes which lead to the status
# of this object. For example it is the "CPU" object on a host
MTE_MON_OBJ = "070"
# Contains performance information (which can be used to create graphs from)
MTE_PERFORMANCE = "100"
# Might contain several messages
MTE_MSG_CONTAINER = "101"
# Contains a single status message
MTE_SINGLE_MSG = "102"
# This is a long text label without status
MTE_LONG_TXT = "110"
# This is a short text label without status
MTE_SHORT_TXT = "111"
# Is a "folder" which has no own state, just computed by its children
MTE_VIRTUAL = "199"
# This map converts between the SAP color codes (key values) and the
# nagios state codes and strings
STATE_VALUE_MAP = {
0: (0, "OK"), # GRAY (inactive or no current info available) -> OK
1: (0, "OK"), # GREEN -> OK
2: (1, "WARN"), # YELLOW -> WARNING
3: (2, "CRIT"), # RED -> CRITICAL
}
STATE_LOGWATCH_MAP = ["O", "O", "W", "C"]
# Monitoring objects of these classes are skipped during processing
SKIP_MTCLASSES = [
MTE_VIRTUAL,
MTE_SUMMARY,
MTE_MON_OBJ,
MTE_SHORT_TXT,
MTE_LONG_TXT,
]
MK_CONFDIR = os.getenv("MK_CONFDIR") or "/etc/check_mk"
MK_VARDIR = os.getenv("MK_VARDIR") or "/var/lib/check_mk_agent"
STATE_FILE = MK_VARDIR + "/sap.state"
state_file_changed = False
# #############################################################################
# Settings to be used to connect to the SAP R/3 host.
local_cfg = {
"ashost": "localhost",
"sysnr": "00",
"client": "100",
"user": "",
"passwd": "",
"trace": "3",
"loglevel": "warn",
#'lang': 'EN',
#'host_prefix': 'FOOBAR_',
}
# A list of strings, while the string must match the full path to one or
# several monitor objects. We use unix shell patterns during matching, so
# you can use several chars as placeholders:
#
# * matches everything
# ? matches any single character
# [seq] matches any character in seq
# [!seq] matches any character not in seq
#
# The * matches the whole following string and does not end on next "/".
# For examples, take a look at the default config file (/etc/check_mk/sap.cfg).
monitor_paths = [
"SAP CCMS Monitor Templates/Dialog Overview/*",
]
monitor_types = [] # type: list[str]
config_file = MK_CONFDIR + "/sap.cfg"
cfg = {} # type: list[dict[Any, Any]] | dict[Any, Any]
if os.path.exists(config_file):
with open(config_file) as opened_file:
exec(opened_file.read()) # nosec B102 # BNS:a29406
if isinstance(cfg, dict):
cfg = [cfg]
else:
cfg = [local_cfg]
# Load the state file into memory
try:
with open(STATE_FILE) as opened_file:
states = ast.literal_eval(opened_file.read())
except IOError:
states = {}
# index of all logfiles which have been found in a run. This is used to
# remove logfiles which are not available anymore from the states dict.
logfiles = []
# #############################################################################
#
# HELPERS
#
class SapError(Exception):
pass
def to_be_monitored(path, toplevel_match=False):
for rule in monitor_paths:
if toplevel_match and rule.count("/") > 1:
rule = "/".join(rule.split("/")[:2])
if fnmatch.fnmatch(path, rule):
return True
return False
def node_path(tree, node, path=""):
if path:
path = node["MTNAMESHRT"].rstrip() + SEPARATOR + path
else:
path = node["MTNAMESHRT"].rstrip()
if node["ALPARINTRE"] > 0:
parent_node = tree[node["ALPARINTRE"] - 1]
return node_path(tree, parent_node, path)
return path
#
# API ACCESS FUNCTIONS
#
def login(conn):
f = conn.call(
"BAPI_XMI_LOGON",
EXTCOMPANY="Checkmk GmbH",
EXTPRODUCT="Check_MK SAP Agent",
INTERFACE="XAL",
VERSION="1.0",
)
return f["SESSIONID"]
def logout(conn):
conn.call("BAPI_XMI_LOGOFF", INTERFACE="XAL")
def mon_list(conn, cfg_entry):
f = conn.call("BAPI_SYSTEM_MON_GETLIST", EXTERNAL_USER_NAME=cfg_entry["user"])
l = []
for mon in f["MONITOR_NAMES"]:
l.append((mon["MS_NAME"].rstrip(), mon["MONI_NAME"].rstrip()))
return l
# def ms_list(cfg):
# f = conn.call("BAPI_SYSTEM_MS_GETLIST", EXTERNAL_USER_NAME=cfg["user"])
# l = []
# for ms in f["MONITOR_SETS"]:
# l.append(ms["NAME"].rstrip())
# return l
def mon_tree(conn, cfg_entry, ms_name, mon_name):
f = conn.call(
"BAPI_SYSTEM_MON_GETTREE",
EXTERNAL_USER_NAME=cfg_entry["user"],
MONITOR_NAME={
"MS_NAME": ms_name,
"MONI_NAME": mon_name,
},
)
tree = f["TREE_NODES"]
for node in tree:
try:
node["PATH"] = ms_name + SEPARATOR + node_path(tree, node)
except RecursionError:
raise SapError(
(
"Could not calculate path, recursion limit reached. "
"Reorganise your SAP data to get past this error. "
"Element that causes this: {node}"
).format(node=node)
)
return tree
def tid(node):
return {
"MTSYSID": node["MTSYSID"].strip(),
"MTMCNAME": node["MTMCNAME"].strip(),
"MTNUMRANGE": node["MTNUMRANGE"].strip(),
"MTUID": node["MTUID"].strip(),
"MTCLASS": node["MTCLASS"].strip(),
"MTINDEX": node["MTINDEX"].strip(),
"EXTINDEX": node["EXTINDEX"].strip(),
}
def mon_perfdata(conn, cfg_entry, node):
f = conn.call(
"BAPI_SYSTEM_MTE_GETPERFCURVAL",
EXTERNAL_USER_NAME=cfg_entry["user"],
TID=tid(node),
)
value = f["CURRENT_VALUE"]["LASTPERVAL"]
f = conn.call(
"BAPI_SYSTEM_MTE_GETPERFPROP",
EXTERNAL_USER_NAME=cfg_entry["user"],
TID=tid(node),
)
if f["PROPERTIES"]["DECIMALS"] != 0:
value = (value + 0.0) / 10 ** f["PROPERTIES"]["DECIMALS"]
uom = f["PROPERTIES"]["VALUNIT"].strip()
return value, uom
def mon_msg(conn, cfg_entry, node):
f = conn.call(
"BAPI_SYSTEM_MTE_GETSMVALUE",
EXTERNAL_USER_NAME=cfg_entry["user"],
TID=tid(node),
)
data = f["VALUE"]
dt = parse_dt(data["SMSGDATE"], data["SMSGTIME"])
return (dt, data["MSG"].strip())
def parse_dt(d, t):
d = d.strip()
t = t.strip()
if not d or not t:
return None
return datetime.datetime(*time.strptime(d + t, "%Y%m%d%H%M%S")[:6])
def mon_alerts(conn, cfg_entry, node):
f = conn.call("BAPI_SYSTEM_MTE_GETALERTS", EXTERNAL_USER_NAME=cfg_entry["user"], TID=tid(node))
return f["ALERTS"]
def aid(alert):
return {
"ALSYSID": alert["ALSYSID"],
"MSEGNAME": alert["MSEGNAME"],
"ALUNIQNUM": alert["ALUNIQNUM"],
"ALINDEX": alert["ALINDEX"],
"ALERTDATE": alert["ALERTDATE"],
"ALERTTIME": alert["ALERTTIME"],
}
def alert_details(conn, cfg_entry, alert):
f = conn.call(
"BAPI_SYSTEM_ALERT_GETDETAILS",
EXTERNAL_USER_NAME=cfg_entry["user"],
AID=aid(alert),
)
# prop = f["PROPERTIES"]
state = f["VALUE"]
msg = f["XMI_EXT_MSG"]["MSG"].strip()
return state, msg
def process_alerts(conn, cfg_entry, logs, ms_name, mon_name, node, alerts):
global state_file_changed
sid = node["MTSYSID"].strip() or "Other"
context = node["MTMCNAME"].strip() or "Other"
path = node["PATH"]
# Use the sid as hostname for the logs
hostname = sid
logfile = context + "/" + path
logfiles.append((hostname, logfile))
logs.setdefault(sid, {})
logs[hostname][logfile] = []
newest_log_dt = None
for alert in alerts:
dt = parse_dt(alert["ALERTDATE"], alert["ALERTTIME"])
if (hostname, logfile) in states and states[(hostname, logfile)] >= dt:
continue # skip log messages which are older than the last cached date
if not newest_log_dt or dt > newest_log_dt:
newest_log_dt = dt # store the newest log of this run
alert_state, alert_msg = alert_details(conn, cfg_entry, alert)
# Format lines to "logwatch" format
logs[hostname][logfile].append(
"%s %s %s"
% (
STATE_LOGWATCH_MAP[alert_state["VALUE"]],
dt.strftime("%Y-%m-%d %H:%M:%S"),
alert_msg,
)
)
if newest_log_dt:
# Write newest log age to cache to prevent double processing of logs
states[(hostname, logfile)] = newest_log_dt
state_file_changed = True
return logs
def check(pyrfc, cfg_entry): # pylint: disable=too-many-branches
conn = pyrfc.Connection(
ashost=cfg_entry["ashost"],
sysnr=cfg_entry["sysnr"],
client=cfg_entry["client"],
user=cfg_entry["user"],
passwd=cfg_entry["passwd"],
loglevel=cfg_entry["loglevel"],
lang=cfg_entry["lang"],
)
login(conn)
logs = {} # type: dict[str, dict[str, list]]
sap_data = {} # type: dict[str, list]
# This loop is used to collect all information from SAP
for ms_name, mon_name in mon_list(conn, cfg_entry):
path = ms_name + SEPARATOR + mon_name
if not to_be_monitored(path, True):
continue
tree = mon_tree(conn, cfg_entry, ms_name, mon_name)
for node in tree:
if not to_be_monitored(node["PATH"]):
continue
# sys.stdout.write("%s\n" % node["PATH"])
status_details = "" # type: str | tuple[str, Any]
perfvalue = "-"
uom = "-"
# Use precalculated states
state = {
"VALUE": node["ACTUALVAL"],
"SEVERITY": node["ACTUALSEV"],
}
if state["VALUE"] not in STATE_VALUE_MAP:
sys.stdout.write("UNHANDLED STATE VALUE\n")
sys.exit(1)
#
# Handle different object classes individually
# to get details about them
#
if monitor_types and node["MTCLASS"] not in monitor_types:
continue # Skip unwanted classes if class filtering is enabled
if node["MTCLASS"] == MTE_PERFORMANCE:
perfvalue, this_uom = mon_perfdata(conn, cfg_entry, node)
uom = this_uom if this_uom else uom
elif node["MTCLASS"] == MTE_SINGLE_MSG:
status_details = "%s: %s" % mon_msg(conn, cfg_entry, node)
elif node["MTCLASS"] == MTE_MSG_CONTAINER:
alerts = mon_alerts(conn, cfg_entry, node)
logs = process_alerts(conn, cfg_entry, logs, ms_name, mon_name, node, alerts)
if len(alerts) > 0:
last_alert = alerts[-1]
dt = parse_dt(last_alert["ALERTDATE"], last_alert["ALERTTIME"])
alert_state, alert_msg = alert_details(conn, cfg_entry, last_alert)
last_msg = "%s: %s - %s" % (
dt,
STATE_VALUE_MAP[alert_state["VALUE"]][1],
alert_msg,
)
status_details = "%d Messages, Last: %s" % (len(alerts), last_msg)
else:
status_details = "The log is empty"
elif node["MTCLASS"] not in SKIP_MTCLASSES:
# Add an error to output on unhandled classes
status_details = "UNHANDLED MTCLASS", node["MTCLASS"]
if node["MTCLASS"] not in SKIP_MTCLASSES:
sid = node["MTSYSID"].strip() or "Other"
context = node["MTMCNAME"].strip() or "Other"
path = node["PATH"]
sap_data.setdefault(sid, [])
sap_data[sid].append(
"%s\t%d\t%3d\t%s\t%s\t%s\t%s"
% (
context,
state["VALUE"],
state["SEVERITY"],
path,
perfvalue,
uom,
status_details,
)
)
for host, host_sap in sap_data.items():
sys.stdout.write("<<<<%s%s>>>>\n" % (cfg_entry.get("host_prefix", ""), host))
sys.stdout.write("<<<sap:sep(9)>>>\n")
sys.stdout.write("%s\n" % "\n".join(host_sap))
sys.stdout.write("<<<<>>>>\n")
for host, host_logs in logs.items():
sys.stdout.write("<<<<%s>>>>\n" % host)
sys.stdout.write("<<<logwatch>>>\n")
for log, lines in host_logs.items():
sys.stdout.write("[[[%s]]]\n" % log)
if lines:
sys.stdout.write("\n".join(lines) + "\n")
sys.stdout.write("<<<<>>>>\n")
logout(conn)
conn.close()
def main(): # pylint: disable=too-many-branches
global state_file_changed
try:
import pyrfc # type: ignore[import]
except ImportError as e:
if "No module named pyrfc" in str(e):
sys.stderr.write("Missing the Python module pyrfc.\n")
sys.exit(1)
else:
raise
# It is possible to configure multiple SAP instances to monitor. Loop them all, but
# do not terminate when one connection failed
processed_all = True
try:
for entry in cfg:
try:
check(pyrfc, entry)
sys.stdout.write("<<<sap_state:sep(9)>>>\n%s\tOK\n" % entry["ashost"])
except SapError as e:
sys.stderr.write("ERROR: %s\n" % e)
sys.stdout.write("<<<sap_state:sep(9)>>>\n%s\t%s\n" % (entry["ashost"], e))
processed_all = False
except Exception as e:
sys.stderr.write("ERROR: Unhandled exception (%s)\n" % e)
sys.stdout.write(
"<<<sap_state:sep(9)>>>\n%s\tUnhandled exception (%s)\n" % (entry["ashost"], e)
)
processed_all = False
# Now check whether or not an old logfile needs to be removed. This can only
# be done this way, when all hosts have been reached. Otherwise the cleanup
# is skipped.
if processed_all:
for key in states.keys():
if key not in logfiles:
state_file_changed = True
del states[key]
# Only write the state file once per run. And only when it has been changed
if state_file_changed:
new_file = STATE_FILE + ".new"
state_fd = os.open(new_file, os.O_WRONLY | os.O_CREAT)
fcntl.flock(state_fd, fcntl.LOCK_EX)
os.write(state_fd, repr(states).encode("utf-8"))
os.close(state_fd)
os.rename(STATE_FILE + ".new", STATE_FILE)
except Exception as e:
sys.stderr.write("ERROR: Unhandled exception (%s)\n" % e)
sys.exit(0)
if __name__ == "__main__":
main()