Merge pull request #3930 from l2isbad/megacli

python megacli plugin

7 files changed, 431 insertions, 3 deletions

diff --git a/conf.d/Makefile.am b/conf.d/Makefile.am
index 7bd3622f65..0026e3f130 100644
--- a/conf.d/Makefile.am
+++ b/conf.d/Makefile.am
@@ -53,6 +53,7 @@ dist_pythonconfig_DATA = \
     python.d/litespeed.conf \
     python.d/logind.conf \
     python.d/mdstat.conf \
+    python.d/megacli.conf \
     python.d/memcached.conf \
     python.d/mongodb.conf \
     python.d/mysql.conf \
@@ -111,6 +112,7 @@ dist_healthconfig_DATA = \
     health.d/isc_dhcpd.conf \
     health.d/lighttpd.conf \
     health.d/mdstat.conf \
+    health.d/megacli.conf \
     health.d/memcached.conf \
     health.d/memory.conf \
     health.d/mongodb.conf \
diff --git a/conf.d/health.d/megacli.conf b/conf.d/health.d/megacli.conf
new file mode 100644
index 0000000000..1881a7be14
--- /dev/null
+++ b/conf.d/health.d/megacli.conf
@@ -0,0 +1,48 @@
+   alarm: adapter_state
+      on: megacli.adapter_degraded
+   units: is degraded
+    lookup: sum -10s
+    every: 10s
+    crit: $this > 0
+    info: adapter state
+      to: sysadmin
+
+   template: bbu_relative_charge
+      on: megacli.bbu_relative_charge
+   units: percent
+    lookup: average -10s
+    every: 10s
+    warn: $this <= (($status >= $WARNING)  ? (85) : (80))
+    crit: $this <= (($status == $CRITICAL)  ? (50) : (40))
+    info: BBU relative state of charge
+      to: sysadmin
+
+   template: bbu_cycle_count
+      on: megacli.bbu_cycle_count
+   units: cycle count
+    lookup: average -10s
+    every: 10s
+    warn: $this >= 100
+    crit: $this >= 500
+    info: BBU cycle count
+      to: sysadmin
+
+    alarm: pd_media_errors
+      on: megacli.pd_media_error
+   units: media errors
+    lookup: sum -10s
+    every: 10s
+    warn: $this > 0
+    delay: down 1m multiplier 2 max 10m
+    info: physical drive media errors
+      to: sysadmin
+
+    alarm: pd_predictive_failures
+      on: megacli.pd_predictive_failure
+   units: predictive failures
+    lookup: sum -10s
+    every: 10s
+    warn: $this > 0
+    delay: down 1m multiplier 2 max 10m
+    info: physical drive predictive failures
+      to: sysadmin
diff --git a/conf.d/python.d/megacli.conf b/conf.d/python.d/megacli.conf
new file mode 100644
index 0000000000..d84078ecb1
--- /dev/null
+++ b/conf.d/python.d/megacli.conf
@@ -0,0 +1,68 @@
+# netdata python.d.plugin configuration for megacli
+#
+# This file is in YaML format. Generally the format is:
+#
+# name: value
+#
+
+# ----------------------------------------------------------------------
+# Global Variables
+# These variables set the defaults for all JOBs, however each JOB
+# may define its own, overriding the defaults.
+
+# update_every sets the default data collection frequency.
+# If unset, the python.d.plugin default is used.
+# update_every: 1
+
+# priority controls the order of charts at the netdata dashboard.
+# Lower numbers move the charts towards the top of the page.
+# If unset, the default for python.d.plugin is used.
+# priority: 60000
+
+# retries sets the number of retries to be made in case of failures.
+# If unset, the default for python.d.plugin is used.
+# Attempts to restore the service are made once every update_every
+# and only if the module has collected values in the past.
+# retries: 60
+
+# autodetection_retry sets the job re-check interval in seconds.
+# The job is not deleted if check fails.
+# Attempts to start the job are made once every autodetection_retry.
+# This feature is disabled by default.
+# autodetection_retry: 0
+
+# ----------------------------------------------------------------------
+# JOBS (data collection sources)
+#
+# The default JOBS share the same *name*. JOBS with the same name
+# are mutually exclusive. Only one of them will be allowed running at
+# any time. This allows autodetection to try several alternatives and
+# pick the one that works.
+#
+# Any number of jobs is supported.
+#
+# All python.d.plugin JOBS (for all its modules) support a set of
+# predefined parameters. These are:
+#
+# job_name:
+#     name: myname            # the JOB's name as it will appear at the
+#                             # dashboard (by default is the job_name)
+#                             # JOBs sharing a name are mutually exclusive
+#     update_every: 1         # the JOB's data collection frequency
+#     priority: 60000         # the JOB's order on the dashboard
+#     retries: 60             # the JOB's number of restoration attempts
+#     autodetection_retry: 0  # the JOB's re-check interval in seconds
+#
+# Additionally to the above, megacli also supports the following:
+#
+#     do_battery: yes/no      # default is no. Battery stats (adds additional call to megacli `megacli -AdpBbuCmd -a0`).
+#
+# ----------------------------------------------------------------------
+
+# IMPORTANT
+# The netdata user needs to be able to be able to sudo the megacli program without password:
+# netdata ALL=(root)       NOPASSWD: /path/to/megacli
+
+
+# uncomment the line below to collect battery statistics
+# do_battery: yes
diff --git a/python.d/Makefile.am b/python.d/Makefile.am
index caf4a6210b..43fcef6c69 100644
--- a/python.d/Makefile.am
+++ b/python.d/Makefile.am
@@ -41,6 +41,7 @@ dist_python_DATA = \
     litespeed.chart.py \
     logind.chart.py \
     mdstat.chart.py \
+    megacli.chart.py \
     memcached.chart.py \
     mongodb.chart.py \
     mysql.chart.py \
diff --git a/python.d/README.md b/python.d/README.md
index 2417fc4302..3a246cfbd1 100644
--- a/python.d/README.md
+++ b/python.d/README.md
@@ -1141,6 +1141,35 @@ No configuration is needed.
 
 ---
 
+# megacli
+
+Module collects adapter, physical drives and battery stats.
+
+**Requirements:**
+ * `netdata` user needs to be able to be able to sudo the `megacli` program without password
+
+To grab stats it executes:
+ * `sudo -n megacli -LDPDInfo -aAll`
+ * `sudo -n megacli -AdpBbuCmd -a0`
+ 
+ 
+It produces:
+
+1. **Adapter State**
+
+2. **Physical Drives Media Errors**
+
+3. **Physical Drives Predictive Failures**
+
+4. **Battery Relative State of Charge**
+
+5. **Battery Cycle Count**
+
+### configuration
+Battery stats disabled by default in the module configuration file.
+
+---
+
 # memcached
 
 Memcached monitoring module. Data grabbed from [stats interface](https://github.com/memcached/memcached/wiki/Commands#stats).
diff --git a/python.d/megacli.chart.py b/python.d/megacli.chart.py
new file mode 100644
index 0000000000..594c4fb17b
--- /dev/null
+++ b/python.d/megacli.chart.py
@@ -0,0 +1,280 @@
+# -*- coding: utf-8 -*-
+# Description: megacli netdata python.d module
+# Author: Ilya Mashchenko (l2isbad)
+# SPDX-License-Identifier: GPL-3.0+
+
+
+import re
+
+from bases.FrameworkServices.ExecutableService import ExecutableService
+from bases.collection import find_binary
+
+
+update_every = 5
+
+
+def adapter_charts(ads):
+    order = [
+        'adapter_degraded',
+    ]
+
+    def dims(ad):
+        return [['adapter_{0}_degraded'.format(a.id), 'adapter {0}'.format(a.id)] for a in ad]
+
+    charts = {
+        'adapter_degraded': {
+            'options': [
+                    None, 'Adapter State', 'is degraded', 'adapter', 'megacli.adapter_degraded', 'line'],
+            'lines': dims(ads)
+            },
+    }
+
+    return order, charts
+
+
+def pd_charts(pds):
+    order = [
+        'pd_media_error',
+        'pd_predictive_failure',
+    ]
+
+    def dims(k, pd):
+        return [['slot_{0}_{1}'.format(p.id, k), 'slot {0}'.format(p.id), 'incremental'] for p in pd]
+
+    charts = {
+        'pd_media_error': {
+            'options': [
+                None, 'Physical Drives Media Errors', 'errors/s', 'pd', 'megacli.pd_media_error', 'line'
+            ],
+            'lines': dims("media_error", pds)},
+        'pd_predictive_failure': {
+            'options': [
+                None, 'Physical Drives Predictive Failures', 'failures/s', 'pd', 'megacli.pd_predictive_failure', 'line'
+            ],
+            'lines': dims("predictive_failure", pds)}
+    }
+
+    return order, charts
+
+
+def battery_charts(bats):
+    order = list()
+    charts = dict()
+
+    for b in bats:
+        order.append('bbu_{0}_relative_charge'.format(b.id))
+        charts.update(
+            {
+                'bbu_{0}_relative_charge'.format(b.id): {
+                    'options': [
+                        None, 'Relative State of Charge', '%', 'battery', 'megacli.bbu_relative_charge', 'line'],
+                    'lines': [
+                        ['bbu_{0}_relative_charge'.format(b.id), 'adapter {0}'.format(b.id)],
+                    ]
+                }
+            }
+        )
+
+    for b in bats:
+        order.append('bbu_{0}_cycle_count'.format(b.id))
+        charts.update(
+            {
+                'bbu_{0}_cycle_count'.format(b.id): {
+                    'options': [
+                        None, 'Cycle Count', 'cycle count', 'battery', 'megacli.bbu_cycle_count', 'line'],
+                    'lines': [
+                        ['bbu_{0}_cycle_count'.format(b.id), 'adapter {0}'.format(b.id)],
+                    ]
+                }
+            }
+        )
+
+    return order, charts
+
+
+RE_ADAPTER = re.compile(
+    r'Adapter #([0-9]+) State\s+: ([a-zA-Z]+)'
+)
+
+RE_VD = re.compile(
+    r'Slot Number: ([0-9]+) Media Error Count: ([0-9]+) Predictive Failure Count: ([0-9]+)'
+)
+
+RE_BATTERY = re.compile(
+    r'BBU Capacity Info for Adapter: ([0-9]+) Relative State of Charge: ([0-9]+) % Cycle Count: ([0-9]+)'
+)
+
+
+def find_adapters(d):
+    keys = ("Adapter #", "State")
+    d = ' '.join(v.strip() for v in d if v.startswith(keys))
+    return [Adapter(*v) for v in RE_ADAPTER.findall(d)]
+
+
+def find_pds(d):
+    keys = ("Slot Number",  "Media Error Count", "Predictive Failure Count")
+    d = ' '.join(v.strip() for v in d if v.startswith(keys))
+    return [PD(*v) for v in RE_VD.findall(d)]
+
+
+def find_batteries(d):
+    keys = ('BBU Capacity Info for Adapter', 'Relative State of Charge', 'Cycle Count')
+    d = ' '.join(v.strip() for v in d if v.strip().startswith(keys))
+    return [Battery(*v) for v in RE_BATTERY.findall(d)]
+
+
+class Adapter:
+    def __init__(self, n, state):
+        self.id = n
+        self.state = int(state == 'Degraded')
+
+    def data(self):
+        return {
+            'adapter_{0}_degraded'.format(self.id): self.state,
+            }
+
+
+class PD:
+    def __init__(self, n, media_err, predict_fail):
+        self.id = n
+        self.media_err = media_err
+        self.predict_fail = predict_fail
+
+    def data(self):
+        return {
+            'slot_{0}_media_error'.format(self.id): self.media_err,
+            'slot_{0}_predictive_failure'.format(self.id): self.predict_fail,
+        }
+
+
+class Battery:
+    def __init__(self, adapt_id, rel_charge, cycle_count):
+        self.id = adapt_id
+        self.rel_charge = rel_charge
+        self.cycle_count = cycle_count
+
+    def data(self):
+        return {
+            'bbu_{0}_relative_charge'.format(self.id): self.rel_charge,
+            'bbu_{0}_cycle_count'.format(self.id): self.cycle_count,
+        }
+
+
+# TODO: hardcoded sudo...
+class Megacli:
+    def __init__(self):
+        self.s = find_binary('sudo')
+        self.m = find_binary('megacli')
+        self.sudo_check = [self.s, '-n', '-v']
+        self.disk_info = [self.s, '-n', self.m, '-LDPDInfo', '-aAll']
+        self.battery_info = [self.s, '-n', self.m, '-AdpBbuCmd', '-a0']
+
+    def __bool__(self):
+        return bool(self.s and self.m)
+
+    def __nonzero__(self):
+        return self.__bool__()
+
+
+class Service(ExecutableService):
+    def __init__(self, configuration=None, name=None):
+        ExecutableService.__init__(self, configuration=configuration, name=name)
+        self.order = list()
+        self.definitions = dict()
+        self.megacli = Megacli()
+        self.do_battery = self.configuration.get('do_battery')
+
+    def check_sudo(self):
+        err = self._get_raw_data(command=self.megacli.sudo_check, stderr=True)
+        if err:
+            self.error(''.join(err))
+            return False
+        return True
+
+    def check_disk_info(self):
+        d = self._get_raw_data(command=self.megacli.disk_info)
+        if not d:
+            return False
+
+        ads = find_adapters(d)
+        pds = find_pds(d)
+
+        if not (ads and pds):
+            self.error('failed to parse "{0}" output'.format(' '.join(self.megacli.disk_info)))
+            return False
+
+        o, c = adapter_charts(ads)
+        self.order.extend(o)
+        self.definitions.update(c)
+
+        o, c = pd_charts(pds)
+        self.order.extend(o)
+        self.definitions.update(c)
+
+        return True
+
+    def check_battery(self):
+        d = self._get_raw_data(command=self.megacli.battery_info)
+        if not d:
+            return False
+
+        bats = find_batteries(d)
+
+        if not bats:
+            self.error('failed to parse "{0}" output'.format(' '.join(self.megacli.battery_info)))
+            return False
+
+        o, c = battery_charts(bats)
+        self.order.extend(o)
+        self.definitions.update(c)
+        return True
+
+    def check(self):
+        if not self.megacli:
+            self.error('can\'t locate "sudo" or "megacli" binary')
+            return None
+
+        if not (self.check_sudo() and self.check_disk_info()):
+            return False
+
+        if self.do_battery:
+            self.do_battery = self.check_battery()
+
+        return True
+
+    def get_data(self):
+        data = dict()
+
+        data.update(self.get_adapter_pd_data())
+
+        if self.do_battery:
+            data.update(self.get_battery_data())
+
+        return data or None
+
+    def get_adapter_pd_data(self):
+        raw = self._get_raw_data(command=self.megacli.disk_info)
+        data = dict()
+
+        if not raw:
+            return data
+
+        for a in find_adapters(raw):
+            data.update(a.data())
+
+        for p in find_pds(raw):
+            data.update(p.data())
+
+        return data
+
+    def get_battery_data(self):
+        raw = self._get_raw_data(command=self.megacli.battery_info)
+        data = dict()
+
+        if not raw:
+            return data
+
+        for b in find_batteries(raw):
+            data.update(b.data())
+
+        return data
diff --git a/python.d/python_modules/bases/FrameworkServices/ExecutableService.py b/python.d/python_modules/bases/FrameworkServices/ExecutableService.py
index 9b2e945e79..625f643234 100644
--- a/python.d/python_modules/bases/FrameworkServices/ExecutableService.py
+++ b/python.d/python_modules/bases/FrameworkServices/ExecutableService.py
@@ -17,15 +17,15 @@ class ExecutableService(SimpleService):
         SimpleService.__init__(self, configuration=configuration, name=name)
         self.command = None
 
-    def _get_raw_data(self, stderr=False):
+    def _get_raw_data(self, stderr=False, command=None):
         """
         Get raw data from executed command
         :return: <list>
         """
         try:
-            p = Popen(self.command, stdout=PIPE, stderr=PIPE)
+            p = Popen(command if command else self.command, stdout=PIPE, stderr=PIPE)
         except Exception as error:
-            self.error('Executing command {command} resulted in error: {error}'.format(command=self.command,
+            self.error('Executing command {command} resulted in error: {error}'.format(command=command or self.command,
                                                                                        error=error))
             return None
         data = list()