Updated version 1.7.0+dfsg from 'upstream/1.7.0+dfsg'

with Debian dir e101ee6549c6786aab4b3f5c4f9bbb7638f17e34

1 files changed, 97 insertions, 70 deletions

diff --git a/python.d/mdstat.chart.py b/python.d/mdstat.chart.py
index 0f7d2b44..ca9aba56 100644
--- a/python.d/mdstat.chart.py
+++ b/python.d/mdstat.chart.py
@@ -2,98 +2,125 @@
 # Description: mdstat netdata python.d module
 # Author: l2isbad
 
+from re import compile as re_compile
+from collections import defaultdict
 from base import SimpleService
-from re import compile
+
 priority = 60000
 retries = 60
 update_every = 1
 
+OPERATIONS = ('check', 'resync', 'reshape', 'recovery', 'finish', 'speed')
+
 
 class Service(SimpleService):
     def __init__(self, configuration=None, name=None):
         SimpleService.__init__(self, configuration=configuration, name=name)
-        self.order = ['agr_health']
-        self.definitions = {'agr_health':
-                                {'options':
-                                     [None, 'Faulty devices in MD', 'failed disks', 'health', 'md.health', 'line'],
-                                 'lines': []}}
-        self.proc_mdstat = '/proc/mdstat'
-        self.regex_disks = compile(r'((?<=\ )[a-zA-Z_0-9]+(?= : active)).*?((?<= \[)[0-9]+)/([0-9]+(?=\] ))')
-        self.regex_status = compile(r'([a-zA-Z_0-9]+)( : active)[^:]*?([a-z]+) = ([0-9.]+(?=%)).*?((?<=finish=)[0-9.]+)min speed=([0-9]+)')
+        self.regex = dict(disks=re_compile(r' (?P<array>[a-zA-Z_0-9]+) : active .+\['
+                                           r'(?P<total_disks>[0-9]+)/'
+                                           r'(?P<inuse_disks>[0-9])\]'),
+                          status=re_compile(r' (?P<array>[a-zA-Z_0-9]+) : active .+ '
+                                            r'(?P<operation>[a-z]+) = '
+                                            r'(?P<operation_status>[0-9.]+).+finish='
+                                            r'(?P<finish>([0-9.]+))min speed='
+                                            r'(?P<speed>[0-9]+)'))
 
     def check(self):
-        raw_data = self._get_raw_data()
-        if not raw_data:
-            self.error('Cant read mdstat data from %s' % (self.proc_mdstat))
-            return False
-        
-        md_list = [md[0] for md in self.regex_disks.findall(raw_data)]
-
-        if not md_list:
-            self.error('No active arrays in %s' % (self.proc_mdstat))
-            return False
-        else:
-            for md in md_list:
-                self.order.append(md)
-                self.order.append(''.join([md, '_status']))
-                self.order.append(''.join([md, '_rate']))
-                self.definitions['agr_health']['lines'].append([''.join([md, '_health']), md, 'absolute'])
-                self.definitions[md] = {'options':
-                                            [None, '%s disks stats' % md, 'disks', md, 'md.disks', 'stacked'],
-                                        'lines': [[''.join([md, '_total']), 'total', 'absolute'],
-                                                  [''.join([md, '_inuse']), 'inuse', 'absolute']]}
-                self.definitions[''.join([md, '_status'])] = {'options':
-                                            [None, '%s current status' % md, 'percent', md, 'md.status', 'line'],
-                                        'lines': [[''.join([md, '_resync']), 'resync', 'absolute', 1, 100],
-                                                  [''.join([md, '_recovery']), 'recovery', 'absolute', 1, 100],
-                                                  [''.join([md, '_reshape']), 'reshape', 'absolute', 1, 100],
-                                                  [''.join([md, '_check']), 'check', 'absolute', 1, 100]]}
-                self.definitions[''.join([md, '_rate'])] = {'options':
-                                            [None, '%s operation status' % md, 'rate', md, 'md.rate', 'line'],
-                                        'lines': [[''.join([md, '_finishin']), 'finish min', 'absolute', 1, 100],
-                                                  [''.join([md, '_rate']), 'megabyte/s', 'absolute', -1, 100]]}
-            self.info('Plugin was started successfully. MDs to monitor %s' % (md_list))
-
-            return True
-
-    def _get_raw_data(self):
+        arrays = find_arrays(self._get_raw_data(), self.regex)
+        if not arrays:
+            self.error('Failed to read data from /proc/mdstat or there is no active arrays')
+            return None
+
+        self.order, self.definitions = create_charts(arrays.keys())
+        return True
+
+    @staticmethod
+    def _get_raw_data():
         """
         Read data from /proc/mdstat
         :return: str
         """
         try:
-            with open(self.proc_mdstat, 'rt') as proc_mdstat:
-                raw_result = proc_mdstat.read()
-        except Exception:
+            with open('/proc/mdstat', 'rt') as proc_mdstat:
+                return proc_mdstat.readlines() or None
+        except (OSError, IOError):
             return None
-        else:
-            raw_result = ' '.join(raw_result.split())
-            return raw_result
 
     def _get_data(self):
         """
         Parse data from _get_raw_data()
         :return: dict
         """
-        raw_mdstat = self._get_raw_data()
-        mdstat_disks = self.regex_disks.findall(raw_mdstat)
-        mdstat_status = self.regex_status.findall(raw_mdstat)
-        to_netdata = {}
-
-        for md in mdstat_disks:
-            to_netdata[''.join([md[0], '_total'])] = int(md[1])
-            to_netdata[''.join([md[0], '_inuse'])] = int(md[2])
-            to_netdata[''.join([md[0], '_health'])] = int(md[1]) - int(md[2])
-            to_netdata[''.join([md[0], '_check'])] = 0
-            to_netdata[''.join([md[0], '_resync'])] = 0
-            to_netdata[''.join([md[0], '_reshape'])] = 0
-            to_netdata[''.join([md[0], '_recovery'])] = 0
-            to_netdata[''.join([md[0], '_finishin'])] = 0
-            to_netdata[''.join([md[0], '_rate'])] = 0
-        
-        for md in mdstat_status:
-            to_netdata[''.join([md[0], '_' + md[2]])] = round(float(md[3]) * 100)
-            to_netdata[''.join([md[0], '_finishin'])] = round(float(md[4]) * 100)
-            to_netdata[''.join([md[0], '_rate'])] = round(float(md[5]) / 1000 * 100)
+        raw_data = self._get_raw_data()
+        arrays = find_arrays(raw_data, self.regex)
+        if not arrays:
+            return None
+
+        to_netdata = dict()
+        for array, values in arrays.items():
+            for key, value in values.items():
+                to_netdata['_'.join([array, key])] = value
 
         return to_netdata
+
+
+def find_arrays(raw_data, regex):
+    if raw_data is None:
+        return None
+    data = defaultdict(str)
+    counter = 1
+
+    for row in (elem.strip() for elem in raw_data):
+        if not row:
+            counter += 1
+            continue
+        data[counter] = ' '.join([data[counter], row])
+
+    arrays = dict()
+    for value in data.values():
+        match = regex['disks'].search(value)
+        if not match:
+            continue
+
+        match = match.groupdict()
+        array = match.pop('array')
+        arrays[array] = match
+        arrays[array]['health'] = int(match['total_disks']) - int(match['inuse_disks'])
+        for operation in OPERATIONS:
+            arrays[array][operation] = 0
+
+        match = regex['status'].search(value)
+        if match:
+            match = match.groupdict()
+            if match['operation'] in OPERATIONS:
+                arrays[array][match['operation']] = float(match['operation_status']) * 100
+                arrays[array]['finish'] = float(match['finish']) * 100
+                arrays[array]['speed'] = float(match['speed']) / 1000 * 100
+
+    return arrays or None
+
+
+def create_charts(arrays):
+    order = ['mdstat_health']
+    definitions = dict(mdstat_health={'options': [None, 'Faulty devices in MD', 'failed disks',
+                                                  'health', 'md.health', 'line'],
+                                      'lines': []})
+    for md in arrays:
+        order.append(md)
+        order.append(md + '_status')
+        order.append(md + '_rate')
+        definitions['mdstat_health']['lines'].append([md + '_health', md, 'absolute'])
+        definitions[md] = {'options': [None, '%s disks stats' % md, 'disks', md, 'md.disks', 'stacked'],
+                           'lines': [[md + '_total_disks', 'total', 'absolute'],
+                                     [md + '_inuse_disks', 'inuse', 'absolute']]}
+        definitions[md + '_status'] = {'options': [None, '%s current status' % md,
+                                                   'percent', md, 'md.status', 'line'],
+                                       'lines': [[md + '_resync', 'resync', 'absolute', 1, 100],
+                                                 [md + '_recovery', 'recovery', 'absolute', 1, 100],
+                                                 [md + '_reshape', 'reshape', 'absolute', 1, 100],
+                                                 [md + '_check', 'check', 'absolute', 1, 100]]}
+        definitions[md + '_rate'] = {'options': [None, '%s operation status' % md,
+                                                 'rate', md, 'md.rate', 'line'],
+                                     'lines': [[md + '_finish', 'finish min', 'absolute', 1, 100],
+                                               [md + '_speed', 'MB/s', 'absolute', -1, 100]]}
+    return order, definitions