Source code for insights.parsers.mdstat

Mdstat - file ``/proc/mdstat``

Represents the information in the ``/proc/mdstat`` file.  Several
examples of possible data containe in the file can be found on the
`MDstat wiki page <>`_.

In particular, the discussion here will focus on initial extraction of information
form lines such as::

    Personalities : [raid1] [raid6] [raid5] [raid4]
    md1 : active raid1 sdb2[1] sda2[0]
          136448 blocks [2/2] [UU]

    md2 : active raid1 sdb3[1] sda3[0]
          129596288 blocks [2/2] [UU]

    md3 : active raid5 sdl1[9] sdk1[8] sdj1[7] sdi1[6] sdh1[5] sdg1[4] sdf1[3] sde1[2] sdd1[1] sdc1[0]
          1318680576 blocks level 5, 1024k chunk, algorithm 2 [10/10] [UUUUUUUUUU]

The data contained in ``mdstat`` is represented with three top level members -
``personalities``, ``components`` and ``mds``.


    >>> mdstat = shared[Mdstat]
    >>> mdstat.personalities
    ['raid1', 'raid6', 'raid5', 'raid4']
    >>> len(mdstat.components) # The individual component devices
    >>> mdstat.components[0]['device_name']
    >>> sdb2 = mdstat.components[0]
    >>> sdb2['component_name']
    >>> sdb2['active']
    >>> sdb2['raid']
    >>> sdb2['role']
    >>> sdb2['up']
    >>> sorted(mdstat.mds.keys()) # dictionary of MD devices by device name
    ['md1', 'md2', 'md3']
    >>> mdstat.mds['md1']['active']
    >>> len(mdstat.mds['md1']['devices']) # list of devices in this MD
    >>> mdstat.mds['md1']['devices'][0]['component_name'] # device information
import re
from .. import parser, CommandParser
from insights.specs import Specs

[docs]@parser(Specs.mdstat) class Mdstat(CommandParser): """ Represents the information in the ``/proc/mdstat`` file. Attributes ---------- personalities : list A list of RAID levels the kernel currently supports components : list of dicts A list containing a dict of md component device information Each of these dicts contains the following keys - ``device_name`` : string - name of the array device - ``active`` : boolean - ``True`` if the array is active, ``False`` if it is inactive. - ``component_name`` : string - name of the component device - ``raid`` : string - with the raid level, e.g., "raid1" for "md1" - ``role`` : int - raid role number - ``device_flag`` : str - device component status flag. Known values include 'F' (failed device), 'S', and 'W' - ``up`` : boolean - ``True`` if the component device is up - ``auto_read_only`` : boolean - ``True`` if the array device is "auto-read-only" - ``blocks`` : the number of blocks in the device - ``level`` : the current RAID level, if found in the status line - ``chunk`` : the device chunk size, if found in the status line - ``algorithm`` : the current conflict resolution algorithm, if found in the status line mds : dict of dicts A dictionary keyed on the MD device name, with the following keys - ``name``: Name of the MD device - ``active``: Whether the MD device is active - ``raid``: The RAID type string - ``devices``: a list of the devices in this - ``blocks``, ``level``, ``chunk`` and ``algorithm`` - the same information given above per component device (if found) """
[docs] def parse_content(self, content): self.mds = {} self.components = [] self.personalities = [] current_components = None in_component = False for line in content: line = line.strip() if line.startswith('Personalities'): in_component = False self.personalities = parse_personalities(line) elif line.startswith("md"): # Starting a component array stanza in_component = True current_components = parse_array_start(line) elif not line: # blank line, ending a component array stanza if in_component: self.components.extend(current_components) current_components = None in_component = False else: if in_component: parse_array_status(line, current_components) upstring = parse_upstring(line) if upstring: apply_upstring(upstring, current_components) # Map component devices into MDs dictionary by device name for comp in self.components: devname = comp['device_name'] if devname not in self.mds: # md2 : active raid1 sdb3[1] sda3[0] # 129596288 blocks [2/2] [UU] self.mds[devname] = { 'name': devname, 'active': comp['active'], 'raid': comp['raid'], 'blocks': comp['blocks'], 'devices': [] } for opt in ['level', 'chunk', 'algorithm']: if opt in comp: self.mds[devname][opt] = comp[opt] self.mds[devname]['devices'].append(dict( (k, comp[k]) for k in comp if k in ['component_name', 'role', 'up'] )) # Keep just for backwards compat = { 'personalities': self.personalities, 'components': self.components }
[docs]def parse_personalities(personalities_line): """Parse the "personalities" line of ``/proc/mdstat``. Lines are expected to be like: Personalities : [linear] [raid0] [raid1] [raid5] [raid4] [raid6] If they do not have this format, an error will be raised since it would be considered an unexpected parsing error. Parameters ---------- personalities_line : str A single "Personalities" line from an ``/proc/mdstat`` files. Returns ------- A list of raid "personalities" listed on the line. """ tokens = personalities_line.split() assert tokens.pop(0) == "Personalities" assert tokens.pop(0) == ":" personalities = [] for token in tokens: assert token.startswith('[') and token.endswith(']') personalities.append(token.strip('[]')) return personalities
[docs]def parse_array_start(md_line): """Parse the initial line of a device array stanza in ``/proc/mdstat``. Lines are expected to be like: md2 : active raid1 sdb3[1] sda3[0] If they do not have this format, an error will be raised since it would be considered an unexpected parsing error. Parameters ---------- md_line : str A single line from the start of a device array stanza from a ``/proc/mdstat`` file. Returns ------- A list of dictionaries, one dictionrary for each component device making up the array. """ components = [] tokens = md_line.split() device_name = tokens.pop(0) assert device_name.startswith("md") assert tokens.pop(0) == ":" active_string = tokens.pop(0) active = False if active_string == "active": active = True else: assert active_string == "inactive" raid_read_only_token = tokens.pop(0) auto_read_only = False raid = None if raid_read_only_token == "(auto-read-only)": auto_read_only = True raid = tokens.pop(0) elif active: raid = raid_read_only_token else: # Inactive devices don't list the raid type raid = None tokens.insert(0, raid_read_only_token) for token in tokens: subtokens = re.split(r"\[|]", token) assert len(subtokens) > 1 comp_name = subtokens[0] assert comp_name role = int(subtokens[1]) device_flag = None if len(subtokens) > 2: device_flag = subtokens[2] if device_flag: device_flag = device_flag.strip('()') component_row = {"device_name": device_name, "raid": raid, "active": active, "auto_read_only": auto_read_only, "component_name": comp_name, "role": role, "device_flag": device_flag} components.append(component_row) return components
[docs]def parse_array_status(line, components): """ Parse the array status line, e.g.:: 1318680576 blocks level 5, 1024k chunk, algorithm 2 [10/10] [UUUUUUUUUU] This retrieves the following pieces of information: * ``blocks`` - (int) number of blocks in the whole MD device (always present) * ``level`` - (int) if found, the present RAID level * ``chunksize`` - (str) if found, the size of the data chunk in kilobytes * ``algorithm`` - (int) if found, the current algorithm in use. Because of the way data is stored per-component and not per-array, this then puts the above keys into each of the component dictionaries in the list we've been given. Sample data:: 1250241792 blocks super 1.2 level 5, 64k chunk, algorithm 2 [5/5] [UUUUUU] 1465151808 blocks level 5, 64k chunk, algorithm 2 [4/3] [UUU_] 136448 blocks [2/2] [UU] 6306 blocks super external:imsm<Paste> """ status_line_re = r'(?P<blocks>\d+) blocks' + \ r'(?: super (?P<super>\S+))?' + \ r'(?: level (?P<level>\d+),)?' + \ r'(?: (?P<chunk>\d+k) chunk,)?' + \ r'(?: algorithm (?P<algorithm>\d+))?' # Since we're only called once per line, unless there's a good way to # cache this regular expression in compiled form we're going to have to # compile it each time. status_line_rex = re.compile(status_line_re) match = if match: attributes = {'blocks': int('blocks'))} if'level'): attributes['level'] = int('level')) if'chunk'): attributes['chunk'] ='chunk') if'algorithm'): attributes['algorithm'] = int('algorithm')) for comp in components: comp.update(attributes)
[docs]def parse_upstring(line): """ Parse the subsequent lines of a device array stanza in ``/proc/mdstat`` for the "up" indictor string. Lines are expected to be like: 129596288 blocks [2/2] [UU] or 1318680576 blocks level 5, 1024k chunk, algorithm 2 [10/10] [UUU_UUUUUU] In particular, this method searchs for the string like ``[UU]`` which indicates whether component devices or up, ``U`` or down, ``_``. Parameters ---------- line : str A single line from a device array stanza. Returns ------- The string containing a series of ``U`` and ``\_`` characters if found in the string, and ``None`` if the uptime string is not found. """ UP_STRING_REGEX = r"\[(U|_)+]" match =, line) if match: return'[]') else: return None
[docs]def apply_upstring(upstring, component_list): """Update the dictionaries resulting from ``parse_array_start`` with the "up" key based on the upstring returned from ``parse_upstring``. The function assumes that the upstring and component_list parameters passed in are from the same device array stanza of a ``/proc/mdstat`` file. The function modifies component_list in place, adding or updating the value of the "up" key to True if there is a corresponding ``U`` in the upstring string, or to False if there is a corresponding ``_``. If there the number of rows in component_list does not match the number of characters in upstring, an ``AssertionError`` is raised. Parameters ---------- upstring : str String sequence of ``U``s and ``_``s as determined by the ``parse_upstring`` method component_list : list List of dictionaries output from the ``parse_array_start`` method. """ assert len(upstring) == len(component_list) def add_up_key(comp_dict, up_indicator): assert up_indicator == 'U' or up_indicator == "_" comp_dict['up'] = up_indicator == 'U' for comp_dict, up_indicator in zip(component_list, upstring): add_up_key(comp_dict, up_indicator)