saltstack对递归依赖条件(死循环依赖)的处理
本文将对saltstack中状态文件中require条件产生死循环的情形进行简单的测试和分析
测试思路：
写一个包含递归依赖条件的状态文件，进行测试：
A依赖于B
B依赖于C
C依赖于A
测试过程：
测试的状态文件
salt-master:/srv/salt/test # cat test.sls
test-require-a:
cmd.run:
- name: echo a
- require:
- cmd: test-require-b test-require-b:
cmd.run:
- name: echo b
- require:
- cmd: test-require-c test-require-c:
cmd.run:
- name: echo c
- require:
- cmd: test-require-a salt-master:/srv/salt/test #
测试结果
salt-master:/srv/salt/test # salt 'test248*' state.sls test.test
test248.office.test.jianfeitech.com:
----------
ID: test-require-c
Function: cmd.run
Name: echo c
Result: False
Comment: Recursive requisite found
Changes:
----------
ID: test-require-b
Function: cmd.run
Name: echo b
Result: False
Comment: One or more requisite failed: test.test.test-require-c
Changes:
----------
ID: test-require-a
Function: cmd.run
Name: echo a
Result: False
Comment: One or more requisite failed: test.test.test-require-b
Changes: Summary
------------
Succeeded: 0
Failed: 3
------------
Total: 3
salt-master:/srv/salt/test #
源码：
对状态文件的依赖进行检测处理的代码位于salt.state.py中的call_chunk函数
call_chunk(self, low, running, chunks)
def call_chunk(self, low, running, chunks):
'''
Check if a chunk has any requires, execute the requires and then
the chunk
'''
low = self._mod_aggregate(low, running, chunks)
self._mod_init(low)
tag = _gen_tag(low)
if not low.get('prerequired'):
self.active.add(tag)
requisites = ['require', 'watch', 'prereq', 'onfail', 'onchanges']
if not low.get('__prereq__'):
requisites.append('prerequired')
status, reqs = self.check_requisite(low, running, chunks, True)
else:
status, reqs = self.check_requisite(low, running, chunks)
if status == 'unmet':
lost = {}
reqs = []
for requisite in requisites:
lost[requisite] = []
if requisite not in low:
continue
for req in low[requisite]:
req = trim_req(req)
found = False
for chunk in chunks:
req_key = next(iter(req))
req_val = req[req_key]
if req_val is None:
continue
if req_key == 'sls':
# Allow requisite tracking of entire sls files
if fnmatch.fnmatch(chunk['__sls__'], req_val):
if requisite == 'prereq':
chunk['__prereq__'] = True
reqs.append(chunk)
found = True
continue
if (fnmatch.fnmatch(chunk['name'], req_val) or
fnmatch.fnmatch(chunk['__id__'], req_val)):
if chunk['state'] == req_key:
if requisite == 'prereq':
chunk['__prereq__'] = True
elif requisite == 'prerequired':
chunk['__prerequired__'] = True
reqs.append(chunk)
found = True
if not found:
lost[requisite].append(req)
if lost['require'] or lost['watch'] or lost['prereq'] or lost['onfail'] or lost['onchanges'] or lost.get('prerequired'):
comment = 'The following requisites were not found:\n'
for requisite, lreqs in six.iteritems(lost):
if not lreqs:
continue
comment += \
'{0}{1}:\n'.format(' ' * 19, requisite)
for lreq in lreqs:
req_key = next(iter(lreq))
req_val = lreq[req_key]
comment += \
'{0}{1}: {2}\n'.format(' ' * 23, req_key, req_val)
running[tag] = {'changes': {},
'result': False,
'comment': comment,
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running
for chunk in reqs:
# Check to see if the chunk has been run, only run it if
# it has not been run already
ctag = _gen_tag(chunk)
if ctag not in running:
if ctag in self.active:
if chunk.get('__prerequired__'):
# Prereq recusive, run this chunk with prereq on
if tag not in self.pre:
low['__prereq__'] = True
self.pre[ctag] = self.call(low, chunks, running)
return running
else:
return running
elif ctag not in running:
log.error('Recursive requisite found')
running[tag] = {
'changes': {},
'result': False,
'comment': 'Recursive requisite found',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running
running = self.call_chunk(chunk, running, chunks)
if self.check_failhard(chunk, running):
running['__FAILHARD__'] = True
return running
if low.get('__prereq__'):
status, reqs = self.check_requisite(low, running, chunks)
self.pre[tag] = self.call(low, chunks, running)
if not self.pre[tag]['changes'] and status == 'change':
self.pre[tag]['changes'] = {'watch': 'watch'}
self.pre[tag]['result'] = None
else:
running = self.call_chunk(low, running, chunks)
if self.check_failhard(chunk, running):
running['__FAILHARD__'] = True
return running
elif status == 'met':
if low.get('__prereq__'):
self.pre[tag] = self.call(low, chunks, running)
else:
running[tag] = self.call(low, chunks, running)
elif status == 'fail':
# if the requisite that failed was due to a prereq on this low state
# show the normal error
if tag in self.pre:
running[tag] = self.pre[tag]
running[tag]['__run_num__'] = self.__run_num
running[tag]['__sls__'] = low['__sls__']
# otherwise the failure was due to a requisite down the chain
else:
# determine what the requisite failures where, and return
# a nice error message
failed_requisites = set()
# look at all requisite types for a failure
for req_lows in six.itervalues(reqs):
for req_low in req_lows:
req_tag = _gen_tag(req_low)
req_ret = self.pre.get(req_tag, running.get(req_tag))
# if there is no run output for the requisite it
# can't be the failure
if req_ret is None:
continue
# If the result was False (not None) it was a failure
if req_ret['result'] is False:
# use SLS.ID for the key-- so its easier to find
key = '{sls}.{_id}'.format(sls=req_low['__sls__'],
_id=req_low['__id__'])
failed_requisites.add(key) _cmt = 'One or more requisite failed: {0}'.format(
', '.join(str(i) for i in failed_requisites)
)
running[tag] = {
'changes': {},
'result': False,
'comment': _cmt,
'__run_num__': self.__run_num,
'__sls__': low['__sls__']
}
self.__run_num += 1
elif status == 'change' and not low.get('__prereq__'):
ret = self.call(low, chunks, running)
if not ret['changes'] and not ret.get('skip_watch', False):
low = low.copy()
low['sfun'] = low['fun']
low['fun'] = 'mod_watch'
low['__reqs__'] = reqs
ret = self.call(low, chunks, running)
running[tag] = ret
elif status == 'pre':
pre_ret = {'changes': {},
'result': True,
'comment': 'No changes detected',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
running[tag] = pre_ret
self.pre[tag] = pre_ret
self.__run_num += 1
elif status == 'onfail':
running[tag] = {'changes': {},
'result': True,
'comment': 'State was not run because onfail req did not change',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
elif status == 'onchanges':
running[tag] = {'changes': {},
'result': True,
'comment': 'State was not run because none of the onchanges reqs changed',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
else:
if low.get('__prereq__'):
self.pre[tag] = self.call(low, chunks, running)
else:
running[tag] = self.call(low, chunks, running)
if tag in running:
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running
此外verify_data函数中也会对数据的依赖做检测，代码如下：
def verify_data(self, data):
'''
Verify the data, return an error statement if something is wrong
'''
errors = []
if 'state' not in data:
errors.append('Missing "state" data')
if 'fun' not in data:
errors.append('Missing "fun" data')
if 'name' not in data:
errors.append('Missing "name" data')
if data['name'] and not isinstance(data['name'], six.string_types):
errors.append(
'ID {0!r} in SLS {1!r} is not formed as a string, but is '
'a {2}'.format(
data['name'], data['__sls__'], type(data['name']).__name__)
)
if errors:
return errors
full = data['state'] + '.' + data['fun']
if full not in self.states:
if '__sls__' in data:
errors.append(
'State \'{0}\' was not found in SLS \'{1}\''.format(
full,
data['__sls__']
)
)
reason = self.states.missing_fun_string(full)
if reason:
errors.append('Reason: {0}'.format(reason))
else:
errors.append(
'Specified state \'{0}\' was not found'.format(
full
)
)
else:
# First verify that the parameters are met
aspec = salt.utils.args.get_function_argspec(self.states[full])
arglen = 0
deflen = 0
if isinstance(aspec.args, list):
arglen = len(aspec.args)
if isinstance(aspec.defaults, tuple):
deflen = len(aspec.defaults)
for ind in range(arglen - deflen):
if aspec.args[ind] not in data:
errors.append(
'Missing parameter {0} for state {1}'.format(
aspec.args[ind],
full
)
)
# If this chunk has a recursive require, then it will cause a
# recursive loop when executing, check for it
reqdec = ''
if 'require' in data:
reqdec = 'require'
if 'watch' in data:
# Check to see if the service has a mod_watch function, if it does
# not, then just require
# to just require extend the require statement with the contents
# of watch so that the mod_watch function is not called and the
# requisite capability is still used
if '{0}.mod_watch'.format(data['state']) not in self.states:
if 'require' in data:
data['require'].extend(data.pop('watch'))
else:
data['require'] = data.pop('watch')
reqdec = 'require'
else:
reqdec = 'watch'
if reqdec:
for req in data[reqdec]:
reqfirst = next(iter(req))
if data['state'] == reqfirst:
if (fnmatch.fnmatch(data['name'], req[reqfirst])
or fnmatch.fnmatch(data['__id__'], req[reqfirst])):
err = ('Recursive require detected in SLS {0} for'
' require {1} in ID {2}').format(
data['__sls__'],
req,
data['__id__'])
errors.append(err)
return errors
完整的state.py如下：
# -*- coding: utf-8 -*-
'''
The module used to execute states in salt. A state is unlike a module
execution in that instead of just executing a command it ensure that a
certain state is present on the system. The data sent to the state calls is as follows:
{ 'state': '<state module name>',
'fun': '<state function name>',
'name': '<the name argument passed to all states>'
'argn': '<arbitrary argument, can have many of these>'
}
''' # Import python libs
from __future__ import absolute_import
import os
import sys
import copy
import site
import fnmatch
import logging
import datetime
import traceback
import re # Import salt libs
import salt.utils
import salt.loader
import salt.minion
import salt.pillar
import salt.fileclient
import salt.utils.event
import salt.utils.url
import salt.syspaths as syspaths
from salt.utils import context, immutabletypes
from salt.template import compile_template, compile_template_str
from salt.exceptions import SaltRenderError, SaltReqTimeoutError, SaltException
from salt.utils.odict import OrderedDict, DefaultOrderedDict # Import third party libs
# pylint: disable=import-error,no-name-in-module,redefined-builtin
import salt.ext.six as six
from salt.ext.six.moves import range
# pylint: enable=import-error,no-name-in-module,redefined-builtin log = logging.getLogger(__name__) # These are keywords passed to state module functions which are to be used
# by salt in this state module and not on the actual state module function
STATE_REQUISITE_KEYWORDS = frozenset([
'onchanges',
'onfail',
'prereq',
'prerequired',
'watch',
'require',
'listen',
])
STATE_REQUISITE_IN_KEYWORDS = frozenset([
'onchanges_in',
'onfail_in',
'prereq_in',
'watch_in',
'require_in',
'listen_in',
])
STATE_RUNTIME_KEYWORDS = frozenset([
'fun',
'state',
'check_cmd',
'failhard',
'onlyif',
'unless',
'order',
'prereq',
'prereq_in',
'prerequired',
'reload_modules',
'reload_grains',
'reload_pillar',
'fire_event',
'saltenv',
'use',
'use_in',
'__env__',
'__sls__',
'__id__',
'__pub_user',
'__pub_arg',
'__pub_jid',
'__pub_fun',
'__pub_tgt',
'__pub_ret',
'__pub_pid',
'__pub_tgt_type',
'__prereq__',
]) STATE_INTERNAL_KEYWORDS = STATE_REQUISITE_KEYWORDS.union(STATE_REQUISITE_IN_KEYWORDS).union(STATE_RUNTIME_KEYWORDS) def _odict_hashable(self):
return id(self) OrderedDict.__hash__ = _odict_hashable def split_low_tag(tag):
'''
Take a low tag and split it back into the low dict that it came from
'''
state, id_, name, fun = tag.split('_|-') return {'state': state,
'__id__': id_,
'name': name,
'fun': fun} def _gen_tag(low):
'''
Generate the running dict tag string from the low data structure
'''
return '{0[state]}_|-{0[__id__]}_|-{0[name]}_|-{0[fun]}'.format(low) def _l_tag(name, id_):
low = {'name': 'listen_{0}'.format(name),
'__id__': 'listen_{0}'.format(id_),
'state': 'Listen_Error',
'fun': 'Listen_Error'}
return _gen_tag(low) def trim_req(req):
'''
Trim any function off of a requisite
'''
reqfirst = next(iter(req))
if '.' in reqfirst:
return {reqfirst.split('.')[0]: req[reqfirst]}
return req def state_args(id_, state, high):
'''
Return a set of the arguments passed to the named state
'''
args = set()
if id_ not in high:
return args
if state not in high[id_]:
return args
for item in high[id_][state]:
if not isinstance(item, dict):
continue
if len(item) != 1:
continue
args.add(next(iter(item)))
return args def find_name(name, state, high):
'''
Scan high data for the id referencing the given name
'''
ext_id = ''
if name in high:
ext_id = name
else:
# We need to scan for the name
for nid in high:
if state in high[nid]:
if isinstance(
high[nid][state],
list):
for arg in high[nid][state]:
if not isinstance(arg, dict):
continue
if len(arg) != 1:
continue
if arg[next(iter(arg))] == name:
ext_id = nid
return ext_id def format_log(ret):
'''
Format the state into a log message
'''
msg = ''
if isinstance(ret, dict):
# Looks like the ret may be a valid state return
if 'changes' in ret:
# Yep, looks like a valid state return
chg = ret['changes']
if not chg:
if ret['comment']:
msg = ret['comment']
else:
msg = 'No changes made for {0[name]}'.format(ret)
elif isinstance(chg, dict):
if 'diff' in chg:
if isinstance(chg['diff'], six.string_types):
msg = 'File changed:\n{0}'.format(chg['diff'])
if all([isinstance(x, dict) for x in six.itervalues(chg)]):
if all([('old' in x and 'new' in x)
for x in six.itervalues(chg)]):
msg = 'Made the following changes:\n'
for pkg in chg:
old = chg[pkg]['old']
if not old and old not in (False, None):
old = 'absent'
new = chg[pkg]['new']
if not new and new not in (False, None):
new = 'absent'
msg += '{0!r} changed from {1!r} to ' \
'{2!r}\n'.format(pkg, old, new)
if not msg:
msg = str(ret['changes'])
if ret['result'] is True or ret['result'] is None:
log.info(msg)
else:
log.error(msg)
else:
# catch unhandled data
log.info(str(ret)) def master_compile(master_opts, minion_opts, grains, id_, saltenv):
'''
Compile the master side low state data, and build the hidden state file
'''
st_ = MasterHighState(master_opts, minion_opts, grains, id_, saltenv)
return st_.compile_highstate() def ishashable(obj):
try:
hash(obj)
except TypeError:
return False
return True class StateError(Exception):
'''
Custom exception class.
'''
pass class Compiler(object):
'''
Class used to compile and manage the High Data structure
'''
def __init__(self, opts, renderers):
self.opts = opts
self.rend = renderers # We need __setstate__ and __getstate__ to avoid pickling errors since
# 'self.rend' contains a function reference which is not picklable.
# These methods are only used when pickling so will not be used on
# non-Windows platforms.
def __setstate__(self, state):
self.__init__(state['opts']) def __getstate__(self):
return {'opts': self.opts} def render_template(self, template, **kwargs):
'''
Enforce the states in a template
'''
high = compile_template(
template, self.rend, self.opts['renderer'], **kwargs)
if not high:
return high
return self.pad_funcs(high) def pad_funcs(self, high):
'''
Turns dot delimited function refs into function strings
'''
for name in high:
if not isinstance(high[name], dict):
if isinstance(high[name], six.string_types):
# Is this is a short state? It needs to be padded!
if '.' in high[name]:
comps = high[name].split('.')
if len(comps) >= 2:
# Merge the comps
comps[1] = '.'.join(comps[1:len(comps)])
high[name] = {
# '__sls__': template,
# '__env__': None,
comps[0]: [comps[1]]
}
continue
continue
skeys = set()
for key in sorted(high[name]):
if key.startswith('_'):
continue
if not isinstance(high[name][key], list):
continue
if '.' in key:
comps = key.split('.')
if len(comps) >= 2:
# Merge the comps
comps[1] = '.'.join(comps[1:len(comps)])
# Salt doesn't support state files such as:
#
# /etc/redis/redis.conf:
# file.managed:
# - user: redis
# - group: redis
# - mode: 644
# file.comment:
# - regex: ^requirepass
if comps[0] in skeys:
continue
high[name][comps[0]] = high[name].pop(key)
high[name][comps[0]].append(comps[1])
skeys.add(comps[0])
continue
skeys.add(key)
return high def verify_high(self, high):
'''
Verify that the high data is viable and follows the data structure
'''
errors = []
if not isinstance(high, dict):
errors.append('High data is not a dictionary and is invalid')
reqs = {}
for name, body in six.iteritems(high):
if name.startswith('__'):
continue
if not isinstance(name, six.string_types):
errors.append(
'ID {0!r} in SLS {1!r} is not formed as a string, but is '
'a {2}'.format(name, body['__sls__'], type(name).__name__)
)
if not isinstance(body, dict):
err = ('The type {0} in {1} is not formatted as a dictionary'
.format(name, body))
errors.append(err)
continue
for state in body:
if state.startswith('__'):
continue
if not isinstance(body[state], list):
errors.append(
'State {0!r} in SLS {1!r} is not formed as a list'
.format(name, body['__sls__'])
)
else:
fun = 0
if '.' in state:
fun += 1
for arg in body[state]:
if isinstance(arg, six.string_types):
fun += 1
if ' ' in arg.strip():
errors.append(('The function "{0}" in state '
'"{1}" in SLS "{2}" has '
'whitespace, a function with whitespace is '
'not supported, perhaps this is an argument '
'that is missing a ":"').format(
arg,
name,
body['__sls__']))
elif isinstance(arg, dict):
# The arg is a dict, if the arg is require or
# watch, it must be a list.
#
# Add the requires to the reqs dict and check them
# all for recursive requisites.
argfirst = next(iter(arg))
if argfirst in ('require', 'watch', 'prereq', 'onchanges'):
if not isinstance(arg[argfirst], list):
errors.append(('The {0}'
' statement in state {1!r} in SLS {2!r} '
'needs to be formed as a list').format(
argfirst,
name,
body['__sls__']
))
# It is a list, verify that the members of the
# list are all single key dicts.
else:
reqs[name] = {'state': state}
for req in arg[argfirst]:
if not isinstance(req, dict):
err = ('Requisite declaration {0}'
' in SLS {1} is not formed as a'
' single key dictionary').format(
req,
body['__sls__'])
errors.append(err)
continue
req_key = next(iter(req))
req_val = req[req_key]
if '.' in req_key:
errors.append((
'Invalid requisite type {0!r} '
'in state {1!r}, in SLS '
'{2!r}. Requisite types must '
'not contain dots, did you '
'mean {3!r}?'.format(
req_key,
name,
body['__sls__'],
req_key[:req_key.find('.')]
)
))
if not ishashable(req_val):
errors.append((
'Illegal requisite "{0}", '
'is SLS {1}\n'
).format(
str(req_val),
body['__sls__']))
continue # Check for global recursive requisites
reqs[name][req_val] = req_key
# I am going beyond 80 chars on
# purpose, this is just too much
# of a pain to deal with otherwise
if req_val in reqs:
if name in reqs[req_val]:
if reqs[req_val][name] == state:
if reqs[req_val]['state'] == reqs[name][req_val]:
err = ('A recursive '
'requisite was found, SLS '
'"{0}" ID "{1}" ID "{2}"'
).format(
body['__sls__'],
name,
req_val
)
errors.append(err)
# Make sure that there is only one key in the
# dict
if len(list(arg)) != 1:
errors.append(('Multiple dictionaries '
'defined in argument of state {0!r} in SLS'
' {1!r}').format(
name,
body['__sls__']))
if not fun:
if state == 'require' or state == 'watch':
continue
errors.append(('No function declared in state {0!r} in'
' SLS {1!r}').format(state, body['__sls__']))
elif fun > 1:
errors.append(
'Too many functions declared in state {0!r} in '
'SLS {1!r}'.format(state, body['__sls__'])
)
return errors def order_chunks(self, chunks):
'''
Sort the chunk list verifying that the chunks follow the order
specified in the order options.
'''
cap = 1
for chunk in chunks:
if 'order' in chunk:
if not isinstance(chunk['order'], int):
continue chunk_order = chunk['order']
if chunk_order > cap - 1 and chunk_order > 0:
cap = chunk_order + 100
for chunk in chunks:
if 'order' not in chunk:
chunk['order'] = cap
continue if not isinstance(chunk['order'], (int, float)):
if chunk['order'] == 'last':
chunk['order'] = cap + 1000000
else:
chunk['order'] = cap
if 'name_order' in chunk:
chunk['order'] = chunk['order'] + chunk.pop('name_order') / 10000.0
if chunk['order'] < 0:
chunk['order'] = cap + 1000000 + chunk['order']
chunks.sort(key=lambda chunk: (chunk['order'], '{0[state]}{0[name]}{0[fun]}'.format(chunk)))
return chunks def compile_high_data(self, high):
'''
"Compile" the high data as it is retrieved from the CLI or YAML into
the individual state executor structures
'''
chunks = []
for name, body in six.iteritems(high):
if name.startswith('__'):
continue
for state, run in six.iteritems(body):
funcs = set()
names = set()
if state.startswith('__'):
continue
chunk = {'state': state,
'name': name}
if '__sls__' in body:
chunk['__sls__'] = body['__sls__']
if '__env__' in body:
chunk['__env__'] = body['__env__']
chunk['__id__'] = name
for arg in run:
if isinstance(arg, six.string_types):
funcs.add(arg)
continue
if isinstance(arg, dict):
for key, val in six.iteritems(arg):
if key == 'names':
names.update(val)
continue
else:
chunk.update(arg)
if names:
name_order = 1
for entry in names:
live = copy.deepcopy(chunk)
if isinstance(entry, dict):
low_name = next(six.iterkeys(entry))
live['name'] = low_name
live.update(entry[low_name][0])
else:
live['name'] = entry
live['name_order'] = name_order
name_order = name_order + 1
for fun in funcs:
live['fun'] = fun
chunks.append(live)
else:
live = copy.deepcopy(chunk)
for fun in funcs:
live['fun'] = fun
chunks.append(live)
chunks = self.order_chunks(chunks)
return chunks def apply_exclude(self, high):
'''
Read in the __exclude__ list and remove all excluded objects from the
high data
'''
if '__exclude__' not in high:
return high
ex_sls = set()
ex_id = set()
exclude = high.pop('__exclude__')
for exc in exclude:
if isinstance(exc, str):
# The exclude statement is a string, assume it is an sls
ex_sls.add(exc)
if isinstance(exc, dict):
# Explicitly declared exclude
if len(exc) != 1:
continue
key = next(six.iterkeys(exc))
if key == 'sls':
ex_sls.add(exc['sls'])
elif key == 'id':
ex_id.add(exc['id'])
# Now the excludes have been simplified, use them
if ex_sls:
# There are sls excludes, find the associtaed ids
for name, body in six.iteritems(high):
if name.startswith('__'):
continue
if body.get('__sls__', '') in ex_sls:
ex_id.add(name)
for id_ in ex_id:
if id_ in high:
high.pop(id_)
return high class State(object):
'''
Class used to execute salt states
'''
def __init__(self, opts, pillar=None, jid=None):
if 'grains' not in opts:
opts['grains'] = salt.loader.grains(opts)
self.opts = opts
self._pillar_override = pillar
self.opts['pillar'] = self._gather_pillar()
self.state_con = {}
self.load_modules()
self.active = set()
self.mod_init = set()
self.pre = {}
self.__run_num = 0
self.jid = jid
self.instance_id = str(id(self)) def _gather_pillar(self):
'''
Whenever a state run starts, gather the pillar data fresh
'''
pillar = salt.pillar.get_pillar(
self.opts,
self.opts['grains'],
self.opts['id'],
self.opts['environment'],
pillar=self._pillar_override,
pillarenv=self.opts.get('pillarenv')
)
ret = pillar.compile_pillar()
if self._pillar_override and isinstance(self._pillar_override, dict):
ret.update(self._pillar_override)
return ret def _mod_init(self, low):
'''
Check the module initialization function, if this is the first run
of a state package that has a mod_init function, then execute the
mod_init function in the state module.
'''
# ensure that the module is loaded
try:
self.states['{0}.{1}'.format(low['state'], low['fun'])] # pylint: disable=W0106
except KeyError:
return
minit = '{0}.mod_init'.format(low['state'])
if low['state'] not in self.mod_init:
if minit in self.states._dict:
mret = self.states[minit](low)
if not mret:
return
self.mod_init.add(low['state']) def _mod_aggregate(self, low, running, chunks):
'''
Execute the aggregation systems to runtime modify the low chunk
'''
agg_opt = self.functions['config.option']('state_aggregate')
if low.get('aggregate') is True:
agg_opt = low['aggregate']
if agg_opt is True:
agg_opt = [low['state']]
else:
return low
if low['state'] in agg_opt and not low.get('__agg__'):
agg_fun = '{0}.mod_aggregate'.format(low['state'])
if agg_fun in self.states:
try:
low = self.states[agg_fun](low, chunks, running)
low['__agg__'] = True
except TypeError:
log.error('Failed to execute aggregate for state {0}'.format(low['state']))
return low def _run_check(self, low_data):
'''
Check that unless doesn't return 0, and that onlyif returns a 0.
'''
ret = {'result': False}
cmd_opts = {} if 'shell' in self.opts['grains']:
cmd_opts['shell'] = self.opts['grains'].get('shell')
if 'onlyif' in low_data:
if not isinstance(low_data['onlyif'], list):
low_data_onlyif = [low_data['onlyif']]
else:
low_data_onlyif = low_data['onlyif']
for entry in low_data_onlyif:
cmd = self.functions['cmd.retcode'](
entry, ignore_retcode=True, python_shell=True, **cmd_opts)
log.debug('Last command return code: {0}'.format(cmd))
if cmd != 0 and ret['result'] is False:
ret.update({'comment': 'onlyif execution failed',
'skip_watch': True,
'result': True})
return ret
elif cmd == 0:
ret.update({'comment': 'onlyif execution succeeded', 'result': False})
return ret if 'unless' in low_data:
if not isinstance(low_data['unless'], list):
low_data_unless = [low_data['unless']]
else:
low_data_unless = low_data['unless']
for entry in low_data_unless:
cmd = self.functions['cmd.retcode'](
entry, ignore_retcode=True, python_shell=True, **cmd_opts)
log.debug('Last command return code: {0}'.format(cmd))
if cmd == 0 and ret['result'] is False:
ret.update({'comment': 'unless execution succeeded',
'skip_watch': True,
'result': True})
elif cmd != 0:
ret.update({'comment': 'unless execution failed', 'result': False})
return ret # No reason to stop, return ret
return ret def _run_check_cmd(self, low_data):
'''
Alter the way a successful state run is determined
'''
ret = {'result': False}
cmd_opts = {}
if 'shell' in self.opts['grains']:
cmd_opts['shell'] = self.opts['grains'].get('shell')
for entry in low_data['check_cmd']:
cmd = self.functions['cmd.retcode'](
entry, ignore_retcode=True, python_shell=True, **cmd_opts)
log.debug('Last command return code: {0}'.format(cmd))
if cmd == 0 and ret['result'] is False:
ret.update({'comment': 'check_cmd determined the state succeeded', 'result': True})
elif cmd != 0:
ret.update({'comment': 'check_cmd determined the state failed', 'result': False})
return ret
return ret def load_modules(self, data=None):
'''
Load the modules into the state
'''
log.info('Loading fresh modules for state activity')
self.utils = salt.loader.utils(self.opts)
self.functions = salt.loader.minion_mods(self.opts, self.state_con, utils=self.utils)
if isinstance(data, dict):
if data.get('provider', False):
if isinstance(data['provider'], str):
providers = [{data['state']: data['provider']}]
elif isinstance(data['provider'], list):
providers = data['provider']
else:
providers = {}
for provider in providers:
for mod in provider:
funcs = salt.loader.raw_mod(self.opts,
provider[mod],
self.functions)
if funcs:
for func in funcs:
f_key = '{0}{1}'.format(
mod,
func[func.rindex('.'):]
)
self.functions[f_key] = funcs[func]
self.states = salt.loader.states(self.opts, self.functions, self.utils)
self.rend = salt.loader.render(self.opts, self.functions, states=self.states) def module_refresh(self):
'''
Refresh all the modules
'''
log.debug('Refreshing modules...')
if self.opts['grains'].get('os') != 'MacOS':
# In case a package has been installed into the current python
# process 'site-packages', the 'site' module needs to be reloaded in
# order for the newly installed package to be importable.
try:
reload(site)
except RuntimeError:
log.error('Error encountered during module reload. Modules were not reloaded.')
self.load_modules()
if not self.opts.get('local', False) and self.opts.get('multiprocessing', True):
self.functions['saltutil.refresh_modules']() def check_refresh(self, data, ret):
'''
Check to see if the modules for this state instance need to be updated,
only update if the state is a file or a package and if it changed
something. If the file function is managed check to see if the file is a
possible module type, e.g. a python, pyx, or .so. Always refresh if the
function is recurse, since that can lay down anything.
'''
_reload_modules = False
if data.get('reload_grains', False):
log.debug('Refreshing grains...')
self.opts['grains'] = salt.loader.grains(self.opts)
_reload_modules = True if data.get('reload_pillar', False):
log.debug('Refreshing pillar...')
self.opts['pillar'] = self._gather_pillar()
_reload_modules = True if data.get('reload_modules', False) or _reload_modules:
# User explicitly requests a reload
self.module_refresh()
return if not ret['changes']:
return if data['state'] == 'file':
if data['fun'] == 'managed':
if data['name'].endswith(
('.py', '.pyx', '.pyo', '.pyc', '.so')):
self.module_refresh()
elif data['fun'] == 'recurse':
self.module_refresh()
elif data['fun'] == 'symlink':
if 'bin' in data['name']:
self.module_refresh()
elif data['state'] in ('pkg', 'ports'):
self.module_refresh() def verify_ret(self, ret):
'''
Verify the state return data
'''
if not isinstance(ret, dict):
raise SaltException(
'Malformed state return, return must be a dict'
)
bad = []
for val in ['name', 'result', 'changes', 'comment']:
if val not in ret:
bad.append(val)
if bad:
raise SaltException(
('The following keys were not present in the state '
'return: {0}'
).format(','.join(bad))) def verify_data(self, data):
'''
Verify the data, return an error statement if something is wrong
'''
errors = []
if 'state' not in data:
errors.append('Missing "state" data')
if 'fun' not in data:
errors.append('Missing "fun" data')
if 'name' not in data:
errors.append('Missing "name" data')
if data['name'] and not isinstance(data['name'], six.string_types):
errors.append(
'ID {0!r} in SLS {1!r} is not formed as a string, but is '
'a {2}'.format(
data['name'], data['__sls__'], type(data['name']).__name__)
)
if errors:
return errors
full = data['state'] + '.' + data['fun']
if full not in self.states:
if '__sls__' in data:
errors.append(
'State \'{0}\' was not found in SLS \'{1}\''.format(
full,
data['__sls__']
)
)
reason = self.states.missing_fun_string(full)
if reason:
errors.append('Reason: {0}'.format(reason))
else:
errors.append(
'Specified state \'{0}\' was not found'.format(
full
)
)
else:
# First verify that the parameters are met
aspec = salt.utils.args.get_function_argspec(self.states[full])
arglen = 0
deflen = 0
if isinstance(aspec.args, list):
arglen = len(aspec.args)
if isinstance(aspec.defaults, tuple):
deflen = len(aspec.defaults)
for ind in range(arglen - deflen):
if aspec.args[ind] not in data:
errors.append(
'Missing parameter {0} for state {1}'.format(
aspec.args[ind],
full
)
)
# If this chunk has a recursive require, then it will cause a
# recursive loop when executing, check for it
reqdec = ''
if 'require' in data:
reqdec = 'require'
if 'watch' in data:
# Check to see if the service has a mod_watch function, if it does
# not, then just require
# to just require extend the require statement with the contents
# of watch so that the mod_watch function is not called and the
# requisite capability is still used
if '{0}.mod_watch'.format(data['state']) not in self.states:
if 'require' in data:
data['require'].extend(data.pop('watch'))
else:
data['require'] = data.pop('watch')
reqdec = 'require'
else:
reqdec = 'watch'
if reqdec:
for req in data[reqdec]:
reqfirst = next(iter(req))
if data['state'] == reqfirst:
if (fnmatch.fnmatch(data['name'], req[reqfirst])
or fnmatch.fnmatch(data['__id__'], req[reqfirst])):
err = ('Recursive require detected in SLS {0} for'
' require {1} in ID {2}').format(
data['__sls__'],
req,
data['__id__'])
errors.append(err)
return errors def verify_high(self, high):
'''
Verify that the high data is viable and follows the data structure
'''
errors = []
if not isinstance(high, dict):
errors.append('High data is not a dictionary and is invalid')
reqs = {}
for name, body in six.iteritems(high):
try:
if name.startswith('__'):
continue
except AttributeError:
pass
if not isinstance(name, six.string_types):
errors.append(
'ID {0!r} in SLS {1!r} is not formed as a string, but '
'is a {2}. It may need to be quoted.'.format(
name, body['__sls__'], type(name).__name__)
)
if not isinstance(body, dict):
err = ('The type {0} in {1} is not formatted as a dictionary'
.format(name, body))
errors.append(err)
continue
for state in body:
if state.startswith('__'):
continue
if body[state] is None:
errors.append(
'ID {0!r} in SLS {1!r} contains a short declaration '
'({2}) with a trailing colon. When not passing any '
'arguments to a state, the colon must be omitted.'
.format(name, body['__sls__'], state)
)
continue
if not isinstance(body[state], list):
errors.append(
'State {0!r} in SLS {1!r} is not formed as a list'
.format(name, body['__sls__'])
)
else:
fun = 0
if '.' in state:
fun += 1
for arg in body[state]:
if isinstance(arg, six.string_types):
fun += 1
if ' ' in arg.strip():
errors.append(('The function "{0}" in state '
'"{1}" in SLS "{2}" has '
'whitespace, a function with whitespace is '
'not supported, perhaps this is an argument '
'that is missing a ":"').format(
arg,
name,
body['__sls__']))
elif isinstance(arg, dict):
# The arg is a dict, if the arg is require or
# watch, it must be a list.
#
# Add the requires to the reqs dict and check them
# all for recursive requisites.
argfirst = next(iter(arg))
if argfirst == 'names':
if not isinstance(arg[argfirst], list):
errors.append(
'The \'names\' argument in state '
'{0!r} in SLS {1!r} needs to be '
'formed as a list'
.format(name, body['__sls__'])
)
if argfirst in ('require', 'watch', 'prereq', 'onchanges'):
if not isinstance(arg[argfirst], list):
errors.append(
'The {0} statement in state {1!r} in '
'SLS {2!r} needs to be formed as a '
'list'.format(argfirst,
name,
body['__sls__'])
)
# It is a list, verify that the members of the
# list are all single key dicts.
else:
reqs[name] = {'state': state}
for req in arg[argfirst]:
if not isinstance(req, dict):
err = ('Requisite declaration {0}'
' in SLS {1} is not formed as a'
' single key dictionary').format(
req,
body['__sls__'])
errors.append(err)
continue
req_key = next(iter(req))
req_val = req[req_key]
if '.' in req_key:
errors.append((
'Invalid requisite type {0!r} '
'in state {1!r}, in SLS '
'{2!r}. Requisite types must '
'not contain dots, did you '
'mean {3!r}?'.format(
req_key,
name,
body['__sls__'],
req_key[:req_key.find('.')]
)
))
if not ishashable(req_val):
errors.append((
'Illegal requisite "{0}", '
'please check your syntax.\n'
).format(str(req_val)))
continue # Check for global recursive requisites
reqs[name][req_val] = req_key
# I am going beyond 80 chars on
# purpose, this is just too much
# of a pain to deal with otherwise
if req_val in reqs:
if name in reqs[req_val]:
if reqs[req_val][name] == state:
if reqs[req_val]['state'] == reqs[name][req_val]:
err = ('A recursive '
'requisite was found, SLS '
'"{0}" ID "{1}" ID "{2}"'
).format(
body['__sls__'],
name,
req_val
)
errors.append(err)
# Make sure that there is only one key in the
# dict
if len(list(arg)) != 1:
errors.append(
'Multiple dictionaries defined in '
'argument of state {0!r} in SLS {1!r}'
.format(name, body['__sls__'])
)
if not fun:
if state == 'require' or state == 'watch':
continue
errors.append(
'No function declared in state {0!r} in SLS {1!r}'
.format(state, body['__sls__'])
)
elif fun > 1:
errors.append(
'Too many functions declared in state {0!r} in '
'SLS {1!r}'.format(state, body['__sls__'])
)
return errors def verify_chunks(self, chunks):
'''
Verify the chunks in a list of low data structures
'''
err = []
for chunk in chunks:
err += self.verify_data(chunk)
return err def order_chunks(self, chunks):
'''
Sort the chunk list verifying that the chunks follow the order
specified in the order options.
'''
cap = 1
for chunk in chunks:
if 'order' in chunk:
if not isinstance(chunk['order'], int):
continue chunk_order = chunk['order']
if chunk_order > cap - 1 and chunk_order > 0:
cap = chunk_order + 100
for chunk in chunks:
if 'order' not in chunk:
chunk['order'] = cap
continue if not isinstance(chunk['order'], (int, float)):
if chunk['order'] == 'last':
chunk['order'] = cap + 1000000
else:
chunk['order'] = cap
if 'name_order' in chunk:
chunk['order'] = chunk['order'] + chunk.pop('name_order') / 10000.0
if chunk['order'] < 0:
chunk['order'] = cap + 1000000 + chunk['order']
chunks.sort(key=lambda chunk: (chunk['order'], '{0[state]}{0[name]}{0[fun]}'.format(chunk)))
return chunks def compile_high_data(self, high):
'''
"Compile" the high data as it is retrieved from the CLI or YAML into
the individual state executor structures
'''
chunks = []
for name, body in six.iteritems(high):
if name.startswith('__'):
continue
for state, run in six.iteritems(body):
funcs = set()
names = set()
if state.startswith('__'):
continue
chunk = {'state': state,
'name': name}
if '__sls__' in body:
chunk['__sls__'] = body['__sls__']
if '__env__' in body:
chunk['__env__'] = body['__env__']
chunk['__id__'] = name
for arg in run:
if isinstance(arg, six.string_types):
funcs.add(arg)
continue
if isinstance(arg, dict):
for key, val in six.iteritems(arg):
if key == 'names':
names.update(val)
elif key == 'state':
# Don't pass down a state override
continue
elif (key == 'name' and
not isinstance(val, six.string_types)):
# Invalid name, fall back to ID
chunk[key] = name
else:
chunk[key] = val
if names:
name_order = 1
for entry in names:
live = copy.deepcopy(chunk)
if isinstance(entry, dict):
low_name = next(six.iterkeys(entry))
live['name'] = low_name
live.update(entry[low_name][0])
else:
live['name'] = entry
live['name_order'] = name_order
name_order = name_order + 1
for fun in funcs:
live['fun'] = fun
chunks.append(live)
else:
live = copy.deepcopy(chunk)
for fun in funcs:
live['fun'] = fun
chunks.append(live)
chunks = self.order_chunks(chunks)
return chunks def reconcile_extend(self, high):
'''
Pull the extend data and add it to the respective high data
'''
errors = []
if '__extend__' not in high:
return high, errors
ext = high.pop('__extend__')
for ext_chunk in ext:
for name, body in six.iteritems(ext_chunk):
if name not in high:
state_type = next(
x for x in body if not x.startswith('__')
)
# Check for a matching 'name' override in high data
id_ = find_name(name, state_type, high)
if id_:
name = id_
else:
errors.append(
'Cannot extend ID \'{0}\' in \'{1}:{2}\'. It is not '
'part of the high state.\n'
'This is likely due to a missing include statement '
'or an incorrectly typed ID.\nEnsure that a '
'state with an ID of \'{0}\' is available\nin '
'environment \'{1}\' and to SLS \'{2}\''.format(
name,
body.get('__env__', 'base'),
body.get('__sls__', 'base'))
)
continue
for state, run in six.iteritems(body):
if state.startswith('__'):
continue
if state not in high[name]:
high[name][state] = run
continue
# high[name][state] is extended by run, both are lists
for arg in run:
update = False
for hind in range(len(high[name][state])):
if isinstance(arg, six.string_types) and isinstance(high[name][state][hind], six.string_types):
# replacing the function, replace the index
high[name][state].pop(hind)
high[name][state].insert(hind, arg)
update = True
continue
if isinstance(arg, dict) and isinstance(high[name][state][hind], dict):
# It is an option, make sure the options match
argfirst = next(iter(arg))
if argfirst == next(iter(high[name][state][hind])):
# If argfirst is a requisite then we must merge
# our requisite with that of the target state
if argfirst in STATE_REQUISITE_KEYWORDS:
high[name][state][hind][argfirst].extend(arg[argfirst])
# otherwise, its not a requisite and we are just extending (replacing)
else:
high[name][state][hind] = arg
update = True
if (argfirst == 'name' and
next(iter(high[name][state][hind])) == 'names'):
# If names are overwritten by name use the name
high[name][state][hind] = arg
if not update:
high[name][state].append(arg)
return high, errors def apply_exclude(self, high):
'''
Read in the __exclude__ list and remove all excluded objects from the
high data
'''
if '__exclude__' not in high:
return high
ex_sls = set()
ex_id = set()
exclude = high.pop('__exclude__')
for exc in exclude:
if isinstance(exc, str):
# The exclude statement is a string, assume it is an sls
ex_sls.add(exc)
if isinstance(exc, dict):
# Explicitly declared exclude
if len(exc) != 1:
continue
key = next(six.iterkeys(exc))
if key == 'sls':
ex_sls.add(exc['sls'])
elif key == 'id':
ex_id.add(exc['id'])
# Now the excludes have been simplified, use them
if ex_sls:
# There are sls excludes, find the associated ids
for name, body in six.iteritems(high):
if name.startswith('__'):
continue
sls = body.get('__sls__', '')
if not sls:
continue
for ex_ in ex_sls:
if fnmatch.fnmatch(sls, ex_):
ex_id.add(name)
for id_ in ex_id:
if id_ in high:
high.pop(id_)
return high def requisite_in(self, high):
'''
Extend the data reference with requisite_in arguments
'''
req_in = set([
'require_in',
'watch_in',
'onfail_in',
'onchanges_in',
'use',
'use_in',
'prereq',
'prereq_in',
])
req_in_all = req_in.union(
set([
'require',
'watch',
'onfail',
'onchanges',
]))
extend = {}
errors = []
for id_, body in six.iteritems(high):
if not isinstance(body, dict):
continue
for state, run in six.iteritems(body):
if state.startswith('__'):
continue
for arg in run:
if isinstance(arg, dict):
# It is not a function, verify that the arg is a
# requisite in statement
if len(arg) < 1:
# Empty arg dict
# How did we get this far?
continue
# Split out the components
key = next(iter(arg))
if key not in req_in:
continue
rkey = key.split('_')[0]
items = arg[key]
if isinstance(items, dict):
# Formatted as a single req_in
for _state, name in six.iteritems(items): # Not a use requisite_in
found = False
if name not in extend:
extend[name] = {}
if '.' in _state:
errors.append((
'Invalid requisite in {0}: {1} for '
'{2}, in SLS {3!r}. Requisites must '
'not contain dots, did you mean {4!r}?'
.format(
rkey,
_state,
name,
body['__sls__'],
_state[:_state.find('.')]
)
))
_state = _state.split(".")[0]
if _state not in extend[name]:
extend[name][_state] = []
extend[name]['__env__'] = body['__env__']
extend[name]['__sls__'] = body['__sls__']
for ind in range(len(extend[name][_state])):
if next(iter(
extend[name][_state][ind])) == rkey:
# Extending again
extend[name][_state][ind][rkey].append(
{state: id_}
)
found = True
if found:
continue
# The rkey is not present yet, create it
extend[name][_state].append(
{rkey: [{state: id_}]}
) if isinstance(items, list):
# Formed as a list of requisite additions
for ind in items:
if not isinstance(ind, dict):
# Malformed req_in
continue
if len(ind) < 1:
continue
_state = next(iter(ind))
name = ind[_state]
if '.' in _state:
errors.append((
'Invalid requisite in {0}: {1} for '
'{2}, in SLS {3!r}. Requisites must '
'not contain dots, did you mean {4!r}?'
.format(
rkey,
_state,
name,
body['__sls__'],
_state[:_state.find('.')]
)
))
_state = _state.split(".")[0]
if key == 'prereq_in':
# Add prerequired to origin
if id_ not in extend:
extend[id_] = {}
if state not in extend[id_]:
extend[id_][state] = []
extend[id_][state].append(
{'prerequired': [{_state: name}]}
)
if key == 'prereq':
# Add prerequired to prereqs
ext_id = find_name(name, _state, high)
if not ext_id:
continue
if ext_id not in extend:
extend[ext_id] = {}
if _state not in extend[ext_id]:
extend[ext_id][_state] = []
extend[ext_id][_state].append(
{'prerequired': [{state: id_}]}
)
continue
if key == 'use_in':
# Add the running states args to the
# use_in states
ext_id = find_name(name, _state, high)
if not ext_id:
continue
ext_args = state_args(ext_id, _state, high)
if ext_id not in extend:
extend[ext_id] = {}
if _state not in extend[ext_id]:
extend[ext_id][_state] = []
ignore_args = req_in_all.union(ext_args)
for arg in high[id_][state]:
if not isinstance(arg, dict):
continue
if len(arg) != 1:
continue
if next(iter(arg)) in ignore_args:
continue
# Don't use name or names
if next(six.iterkeys(arg)) == 'name':
continue
if next(six.iterkeys(arg)) == 'names':
continue
extend[ext_id][_state].append(arg)
continue
if key == 'use':
# Add the use state's args to the
# running state
ext_id = find_name(name, _state, high)
if not ext_id:
continue
loc_args = state_args(id_, state, high)
if id_ not in extend:
extend[id_] = {}
if state not in extend[id_]:
extend[id_][state] = []
ignore_args = req_in_all.union(loc_args)
for arg in high[ext_id][_state]:
if not isinstance(arg, dict):
continue
if len(arg) != 1:
continue
if next(iter(arg)) in ignore_args:
continue
# Don't use name or names
if next(six.iterkeys(arg)) == 'name':
continue
if next(six.iterkeys(arg)) == 'names':
continue
extend[id_][state].append(arg)
continue
found = False
if name not in extend:
extend[name] = {}
if _state not in extend[name]:
extend[name][_state] = []
extend[name]['__env__'] = body['__env__']
extend[name]['__sls__'] = body['__sls__']
for ind in range(len(extend[name][_state])):
if next(iter(
extend[name][_state][ind])) == rkey:
# Extending again
extend[name][_state][ind][rkey].append(
{state: id_}
)
found = True
if found:
continue
# The rkey is not present yet, create it
extend[name][_state].append(
{rkey: [{state: id_}]}
)
high['__extend__'] = []
for key, val in six.iteritems(extend):
high['__extend__'].append({key: val})
req_in_high, req_in_errors = self.reconcile_extend(high)
errors.extend(req_in_errors)
return req_in_high, errors def call(self, low, chunks=None, running=None):
'''
Call a state directly with the low data structure, verify data
before processing.
'''
start_time = datetime.datetime.now()
log.info('Running state [{0}] at time {1}'.format(low['name'], start_time.time().isoformat()))
errors = self.verify_data(low)
if errors:
ret = {
'result': False,
'name': low['name'],
'changes': {},
'comment': '',
}
for err in errors:
ret['comment'] += '{0}\n'.format(err)
ret['__run_num__'] = self.__run_num
self.__run_num += 1
format_log(ret)
self.check_refresh(low, ret)
return ret
else:
ret = {'result': False, 'name': low['name'], 'changes': {}} if not low.get('__prereq__'):
log.info(
'Executing state {0[state]}.{0[fun]} for {0[name]}'.format(
low
)
) if 'provider' in low:
self.load_modules(low) state_func_name = '{0[state]}.{0[fun]}'.format(low)
cdata = salt.utils.format_call(
self.states[state_func_name],
low,
initial_ret={'full': state_func_name},
expected_extra_kws=STATE_INTERNAL_KEYWORDS
) inject_globals = {
# Pass a copy of the running dictionary, the low state chunks and
# the current state dictionaries.
# We pass deep copies here because we don't want any misbehaving
# state module to change these at runtime.
'__low__': immutabletypes.freeze(low),
'__running__': immutabletypes.freeze(running) if running else {},
'__instance_id__': self.instance_id,
'__lowstate__': immutabletypes.freeze(chunks) if chunks else {}
} if low.get('__prereq__'):
test = sys.modules[self.states[cdata['full']].__module__].__opts__['test']
sys.modules[self.states[cdata['full']].__module__].__opts__['test'] = True
try:
# Let's get a reference to the salt environment to use within this
# state call.
#
# If the state function accepts an 'env' keyword argument, it
# allows the state to be overridden(we look for that in cdata). If
# that's not found in cdata, we look for what we're being passed in
# the original data, namely, the special dunder __env__. If that's
# not found we default to 'base'
if ('unless' in low and '{0[state]}.mod_run_check'.format(low) not in self.states) or \
('onlyif' in low and '{0[state]}.mod_run_check'.format(low) not in self.states):
ret.update(self._run_check(low)) if 'saltenv' in low:
inject_globals['__env__'] = str(low['saltenv'])
elif isinstance(cdata['kwargs'].get('env', None), six.string_types):
# User is using a deprecated env setting which was parsed by
# format_call.
# We check for a string type since module functions which
# allow setting the OS environ also make use of the "env"
# keyword argument, which is not a string
inject_globals['__env__'] = str(cdata['kwargs']['env'])
elif '__env__' in low:
# The user is passing an alternative environment using __env__
# which is also not the appropriate choice, still, handle it
inject_globals['__env__'] = str(low['__env__'])
else:
# Let's use the default environment
inject_globals['__env__'] = 'base' if 'result' not in ret or ret['result'] is False:
with context.func_globals_inject(self.states[cdata['full']],
**inject_globals):
ret = self.states[cdata['full']](*cdata['args'],
**cdata['kwargs'])
if 'check_cmd' in low and '{0[state]}.mod_run_check_cmd'.format(low) not in self.states:
ret.update(self._run_check_cmd(low))
self.verify_ret(ret)
except Exception:
trb = traceback.format_exc()
# There are a number of possibilities to not have the cdata
# populated with what we might have expected, so just be smart
# enough to not raise another KeyError as the name is easily
# guessable and fallback in all cases to present the real
# exception to the user
if len(cdata['args']) > 0:
name = cdata['args'][0]
elif 'name' in cdata['kwargs']:
name = cdata['kwargs']['name']
else:
name = low.get('name', low.get('__id__'))
ret = {
'result': False,
'name': name,
'changes': {},
'comment': 'An exception occurred in this state: {0}'.format(
trb)
}
finally:
if low.get('__prereq__'):
sys.modules[self.states[cdata['full']].__module__].__opts__[
'test'] = test # If format_call got any warnings, let's show them to the user
if 'warnings' in cdata:
ret.setdefault('warnings', []).extend(cdata['warnings']) if 'provider' in low:
self.load_modules() if low.get('__prereq__'):
low['__prereq__'] = False
return ret ret['__run_num__'] = self.__run_num
self.__run_num += 1
format_log(ret)
self.check_refresh(low, ret)
finish_time = datetime.datetime.now()
ret['start_time'] = start_time.time().isoformat()
delta = (finish_time - start_time)
# duration in milliseconds.microseconds
ret['duration'] = (delta.seconds * 1000000 + delta.microseconds)/1000.0
log.info('Completed state [{0}] at time {1}'.format(low['name'], finish_time.time().isoformat()))
return ret def call_chunks(self, chunks):
'''
Iterate over a list of chunks and call them, checking for requires.
'''
running = {}
for low in chunks:
if '__FAILHARD__' in running:
running.pop('__FAILHARD__')
return running
tag = _gen_tag(low)
if tag not in running:
running = self.call_chunk(low, running, chunks)
if self.check_failhard(low, running):
return running
self.active = set()
return running def check_failhard(self, low, running):
'''
Check if the low data chunk should send a failhard signal
'''
tag = _gen_tag(low)
if (low.get('failhard', False) or self.opts['failhard']
and tag in running):
return not running[tag]['result']
return False def check_requisite(self, low, running, chunks, pre=False):
'''
Look into the running data to check the status of all requisite
states
'''
present = False
# If mod_watch is not available make it a require
if 'watch' in low:
if '{0}.mod_watch'.format(low['state']) not in self.states:
if 'require' in low:
low['require'].extend(low.pop('watch'))
else:
low['require'] = low.pop('watch')
else:
present = True
if 'require' in low:
present = True
if 'prerequired' in low:
present = True
if 'prereq' in low:
present = True
if 'onfail' in low:
present = True
if 'onchanges' in low:
present = True
if not present:
return 'met', ()
reqs = {
'require': [],
'watch': [],
'prereq': [],
'onfail': [],
'onchanges': []}
if pre:
reqs['prerequired'] = []
for r_state in reqs:
if r_state in low and low[r_state] is not None:
for req in low[r_state]:
req = trim_req(req)
found = False
for chunk in chunks:
req_key = next(iter(req))
req_val = req[req_key]
if req_val is None:
continue
if req_key == 'sls':
# Allow requisite tracking of entire sls files
if fnmatch.fnmatch(chunk['__sls__'], req_val):
found = True
reqs[r_state].append(chunk)
continue
if (fnmatch.fnmatch(chunk['name'], req_val) or
fnmatch.fnmatch(chunk['__id__'], req_val)):
if chunk['state'] == req_key:
found = True
reqs[r_state].append(chunk)
if not found:
return 'unmet', ()
fun_stats = set()
for r_state, chunks in six.iteritems(reqs):
if r_state == 'prereq':
run_dict = self.pre
else:
run_dict = running
for chunk in chunks:
tag = _gen_tag(chunk)
if tag not in run_dict:
fun_stats.add('unmet')
continue
if r_state == 'onfail':
if run_dict[tag]['result'] is True:
fun_stats.add('onfail')
continue
else:
if run_dict[tag]['result'] is False:
fun_stats.add('fail')
continue
if r_state == 'onchanges':
if not run_dict[tag]['changes']:
fun_stats.add('onchanges')
else:
fun_stats.add('onchangesmet')
continue
if r_state == 'watch' and run_dict[tag]['changes']:
fun_stats.add('change')
continue
if r_state == 'prereq' and run_dict[tag]['result'] is None:
fun_stats.add('premet')
if r_state == 'prereq' and not run_dict[tag]['result'] is None:
fun_stats.add('pre')
else:
fun_stats.add('met') if 'unmet' in fun_stats:
status = 'unmet'
elif 'fail' in fun_stats:
status = 'fail'
elif 'pre' in fun_stats:
if 'premet' in fun_stats:
status = 'met'
else:
status = 'pre'
elif 'onfail' in fun_stats:
status = 'onfail'
elif 'onchanges' in fun_stats and 'onchangesmet' not in fun_stats:
status = 'onchanges'
elif 'change' in fun_stats:
status = 'change'
else:
status = 'met' return status, reqs def event(self, chunk_ret, length, fire_event=False):
'''
Fire an event on the master bus If `fire_event` is set to True an event will be sent with the
chunk name in the tag and the chunk result in the event data. If `fire_event` is set to a string such as `mystate/is/finished`,
an event will be sent with the string added to the tag and the chunk
result in the event data. If the `state_events` is set to True in the config, then after the
chunk is evaluated an event will be set up to the master with the
results.
'''
if not self.opts.get('local') and (self.opts.get('state_events', True) or fire_event) and self.opts.get('master_uri'):
ret = {'ret': chunk_ret}
if fire_event is True:
tag = salt.utils.event.tagify(
[self.jid, self.opts['id'], str(chunk_ret['name'])], 'state_result'
)
elif isinstance(fire_event, six.string_types):
tag = salt.utils.event.tagify(
[self.jid, self.opts['id'], str(fire_event)], 'state_result'
)
else:
tag = salt.utils.event.tagify(
[self.jid, 'prog', self.opts['id'], str(chunk_ret['__run_num__'])], 'job'
)
ret['len'] = length
preload = {'jid': self.jid}
self.functions['event.fire_master'](ret, tag, preload=preload) def call_chunk(self, low, running, chunks):
'''
Check if a chunk has any requires, execute the requires and then
the chunk
'''
low = self._mod_aggregate(low, running, chunks)
self._mod_init(low)
tag = _gen_tag(low)
if not low.get('prerequired'):
self.active.add(tag)
requisites = ['require', 'watch', 'prereq', 'onfail', 'onchanges']
if not low.get('__prereq__'):
requisites.append('prerequired')
status, reqs = self.check_requisite(low, running, chunks, True)
else:
status, reqs = self.check_requisite(low, running, chunks)
if status == 'unmet':
lost = {}
reqs = []
for requisite in requisites:
lost[requisite] = []
if requisite not in low:
continue
for req in low[requisite]:
req = trim_req(req)
found = False
for chunk in chunks:
req_key = next(iter(req))
req_val = req[req_key]
if req_val is None:
continue
if req_key == 'sls':
# Allow requisite tracking of entire sls files
if fnmatch.fnmatch(chunk['__sls__'], req_val):
if requisite == 'prereq':
chunk['__prereq__'] = True
reqs.append(chunk)
found = True
continue
if (fnmatch.fnmatch(chunk['name'], req_val) or
fnmatch.fnmatch(chunk['__id__'], req_val)):
if chunk['state'] == req_key:
if requisite == 'prereq':
chunk['__prereq__'] = True
elif requisite == 'prerequired':
chunk['__prerequired__'] = True
reqs.append(chunk)
found = True
if not found:
lost[requisite].append(req)
if lost['require'] or lost['watch'] or lost['prereq'] or lost['onfail'] or lost['onchanges'] or lost.get('prerequired'):
comment = 'The following requisites were not found:\n'
for requisite, lreqs in six.iteritems(lost):
if not lreqs:
continue
comment += \
'{0}{1}:\n'.format(' ' * 19, requisite)
for lreq in lreqs:
req_key = next(iter(lreq))
req_val = lreq[req_key]
comment += \
'{0}{1}: {2}\n'.format(' ' * 23, req_key, req_val)
running[tag] = {'changes': {},
'result': False,
'comment': comment,
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running
for chunk in reqs:
# Check to see if the chunk has been run, only run it if
# it has not been run already
ctag = _gen_tag(chunk)
if ctag not in running:
if ctag in self.active:
if chunk.get('__prerequired__'):
# Prereq recusive, run this chunk with prereq on
if tag not in self.pre:
low['__prereq__'] = True
self.pre[ctag] = self.call(low, chunks, running)
return running
else:
return running
elif ctag not in running:
log.error('Recursive requisite found')
running[tag] = {
'changes': {},
'result': False,
'comment': 'Recursive requisite found',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running
running = self.call_chunk(chunk, running, chunks)
if self.check_failhard(chunk, running):
running['__FAILHARD__'] = True
return running
if low.get('__prereq__'):
status, reqs = self.check_requisite(low, running, chunks)
self.pre[tag] = self.call(low, chunks, running)
if not self.pre[tag]['changes'] and status == 'change':
self.pre[tag]['changes'] = {'watch': 'watch'}
self.pre[tag]['result'] = None
else:
running = self.call_chunk(low, running, chunks)
if self.check_failhard(chunk, running):
running['__FAILHARD__'] = True
return running
elif status == 'met':
if low.get('__prereq__'):
self.pre[tag] = self.call(low, chunks, running)
else:
running[tag] = self.call(low, chunks, running)
elif status == 'fail':
# if the requisite that failed was due to a prereq on this low state
# show the normal error
if tag in self.pre:
running[tag] = self.pre[tag]
running[tag]['__run_num__'] = self.__run_num
running[tag]['__sls__'] = low['__sls__']
# otherwise the failure was due to a requisite down the chain
else:
# determine what the requisite failures where, and return
# a nice error message
failed_requisites = set()
# look at all requisite types for a failure
for req_lows in six.itervalues(reqs):
for req_low in req_lows:
req_tag = _gen_tag(req_low)
req_ret = self.pre.get(req_tag, running.get(req_tag))
# if there is no run output for the requisite it
# can't be the failure
if req_ret is None:
continue
# If the result was False (not None) it was a failure
if req_ret['result'] is False:
# use SLS.ID for the key-- so its easier to find
key = '{sls}.{_id}'.format(sls=req_low['__sls__'],
_id=req_low['__id__'])
failed_requisites.add(key) _cmt = 'One or more requisite failed: {0}'.format(
', '.join(str(i) for i in failed_requisites)
)
running[tag] = {
'changes': {},
'result': False,
'comment': _cmt,
'__run_num__': self.__run_num,
'__sls__': low['__sls__']
}
self.__run_num += 1
elif status == 'change' and not low.get('__prereq__'):
ret = self.call(low, chunks, running)
if not ret['changes'] and not ret.get('skip_watch', False):
low = low.copy()
low['sfun'] = low['fun']
low['fun'] = 'mod_watch'
low['__reqs__'] = reqs
ret = self.call(low, chunks, running)
running[tag] = ret
elif status == 'pre':
pre_ret = {'changes': {},
'result': True,
'comment': 'No changes detected',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
running[tag] = pre_ret
self.pre[tag] = pre_ret
self.__run_num += 1
elif status == 'onfail':
running[tag] = {'changes': {},
'result': True,
'comment': 'State was not run because onfail req did not change',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
elif status == 'onchanges':
running[tag] = {'changes': {},
'result': True,
'comment': 'State was not run because none of the onchanges reqs changed',
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
else:
if low.get('__prereq__'):
self.pre[tag] = self.call(low, chunks, running)
else:
running[tag] = self.call(low, chunks, running)
if tag in running:
self.event(running[tag], len(chunks), fire_event=low.get('fire_event'))
return running def call_listen(self, chunks, running):
'''
Find all of the listen routines and call the associated mod_watch runs
'''
listeners = []
crefs = {}
for chunk in chunks:
crefs[(chunk['state'], chunk['name'])] = chunk
crefs[(chunk['state'], chunk['__id__'])] = chunk
if 'listen' in chunk:
listeners.append({(chunk['state'], chunk['name']): chunk['listen']})
if 'listen_in' in chunk:
for l_in in chunk['listen_in']:
for key, val in six.iteritems(l_in):
listeners.append({(key, val): [{chunk['state']: chunk['name']}]})
mod_watchers = []
errors = {}
for l_dict in listeners:
for key, val in six.iteritems(l_dict):
for listen_to in val:
if not isinstance(listen_to, dict):
continue
for lkey, lval in six.iteritems(listen_to):
if (lkey, lval) not in crefs:
rerror = {_l_tag(lkey, lval):
{
'comment': 'Referenced state {0}: {1} does not exist'.format(lkey, lval),
'name': 'listen_{0}:{1}'.format(lkey, lval),
'result': False,
'changes': {}
}}
errors.update(rerror)
continue
to_tag = _gen_tag(crefs[(lkey, lval)])
if to_tag not in running:
continue
if running[to_tag]['changes']:
if key not in crefs:
rerror = {_l_tag(key[0], key[1]):
{'comment': 'Referenced state {0}: {1} does not exist'.format(key[0], key[1]),
'name': 'listen_{0}:{1}'.format(key[0], key[1]),
'result': False,
'changes': {}}}
errors.update(rerror)
continue
chunk = crefs[key]
low = chunk.copy()
low['sfun'] = chunk['fun']
low['fun'] = 'mod_watch'
low['__id__'] = 'listener_{0}'.format(low['__id__'])
for req in STATE_REQUISITE_KEYWORDS:
if req in low:
low.pop(req)
mod_watchers.append(low)
ret = self.call_chunks(mod_watchers)
running.update(ret)
for err in errors:
errors[err]['__run_num__'] = self.__run_num
self.__run_num += 1
running.update(errors)
return running def call_high(self, high):
'''
Process a high data call and ensure the defined states.
'''
errors = []
# If there is extension data reconcile it
high, ext_errors = self.reconcile_extend(high)
errors += ext_errors
errors += self.verify_high(high)
if errors:
return errors
high, req_in_errors = self.requisite_in(high)
errors += req_in_errors
high = self.apply_exclude(high)
# Verify that the high data is structurally sound
if errors:
return errors
# Compile and verify the raw chunks
chunks = self.compile_high_data(high) # Check for any disabled states
disabled = {}
if 'state_runs_disabled' in self.opts['grains']:
_chunks = copy.deepcopy(chunks)
for low in _chunks:
state_ = '{0}.{1}'.format(low['state'], low['fun'])
for pat in self.opts['grains']['state_runs_disabled']:
if fnmatch.fnmatch(state_, pat):
comment = (
'The state function "{0}" is currently disabled by "{1}", '
'to re-enable, run state.enable {1}.'
).format(
state_,
pat,
)
_tag = _gen_tag(low)
disabled[_tag] = {'changes': {},
'result': False,
'comment': comment,
'__run_num__': self.__run_num,
'__sls__': low['__sls__']}
self.__run_num += 1
chunks.remove(low)
break # If there are extensions in the highstate, process them and update
# the low data chunks
if errors:
return errors
ret = dict(list(disabled.items()) + list(self.call_chunks(chunks).items()))
ret = self.call_listen(chunks, ret) def _cleanup_accumulator_data():
accum_data_path = os.path.join(
salt.utils.get_accumulator_dir(self.opts['cachedir']),
self.instance_id
)
try:
os.remove(accum_data_path)
log.debug('Deleted accumulator data file {0}'.format(
accum_data_path)
)
except OSError:
log.debug('File {0} does not exist, no need to cleanup.'.format(
accum_data_path)
)
_cleanup_accumulator_data() return ret def render_template(self, high, template):
errors = []
if not high:
return high, errors if not isinstance(high, dict):
errors.append(
'Template {0} does not render to a dictionary'.format(template)
)
return high, errors invalid_items = ('include', 'exclude', 'extends')
for item in invalid_items:
if item in high:
errors.append(
'The \'{0}\' declaration found on \'{1}\' is invalid when '
'rendering single templates'.format(item, template)
)
return high, errors for name in high:
if not isinstance(high[name], dict):
if isinstance(high[name], six.string_types):
# Is this is a short state, it needs to be padded
if '.' in high[name]:
comps = high[name].split('.')
high[name] = {
# '__sls__': template,
# '__env__': None,
comps[0]: [comps[1]]
}
continue errors.append(
'ID {0} in template {1} is not a dictionary'.format(
name, template
)
)
continue
skeys = set()
for key in sorted(high[name]):
if key.startswith('_'):
continue
if high[name][key] is None:
errors.append(
'ID {0!r} in template {1} contains a short '
'declaration ({2}) with a trailing colon. When not '
'passing any arguments to a state, the colon must be '
'omitted.'.format(name, template, key)
)
continue
if not isinstance(high[name][key], list):
continue
if '.' in key:
comps = key.split('.')
# Salt doesn't support state files such as:
#
# /etc/redis/redis.conf:
# file.managed:
# - user: redis
# - group: redis
# - mode: 644
# file.comment:
# - regex: ^requirepass
if comps[0] in skeys:
errors.append(
'ID {0!r} in template {1!r} contains multiple '
'state declarations of the same type'
.format(name, template)
)
continue
high[name][comps[0]] = high[name].pop(key)
high[name][comps[0]].append(comps[1])
skeys.add(comps[0])
continue
skeys.add(key) return high, errors def call_template(self, template):
'''
Enforce the states in a template
'''
high = compile_template(
template, self.rend, self.opts['renderer'])
if not high:
return high
high, errors = self.render_template(high, template)
if errors:
return errors
return self.call_high(high) def call_template_str(self, template):
'''
Enforce the states in a template, pass the template as a string
'''
high = compile_template_str(
template, self.rend, self.opts['renderer'])
if not high:
return high
high, errors = self.render_template(high, '<template-str>')
if errors:
return errors
return self.call_high(high) class BaseHighState(object):
'''
The BaseHighState is an abstract base class that is the foundation of
running a highstate, extend it and add a self.state object of type State. When extending this class, please note that ``self.client`` and
``self.matcher`` should be instantiated and handled.
'''
def __init__(self, opts):
self.opts = self.__gen_opts(opts)
self.iorder = 10000
self.avail = self.__gather_avail()
self.serial = salt.payload.Serial(self.opts)
self.building_highstate = {} def __gather_avail(self):
'''
Gather the lists of available sls data from the master
'''
avail = {}
for saltenv in self._get_envs():
avail[saltenv] = self.client.list_states(saltenv)
return avail def __gen_opts(self, opts):
'''
The options used by the High State object are derived from options
on the minion and the master, or just the minion if the high state
call is entirely local.
'''
# If the state is intended to be applied locally, then the local opts
# should have all of the needed data, otherwise overwrite the local
# data items with data from the master
if 'local_state' in opts:
if opts['local_state']:
return opts
mopts = self.client.master_opts()
if not isinstance(mopts, dict):
# An error happened on the master
opts['renderer'] = 'yaml_jinja'
opts['failhard'] = False
opts['state_top'] = salt.utils.url.create('top.sls')
opts['nodegroups'] = {}
opts['file_roots'] = {'base': [syspaths.BASE_FILE_ROOTS_DIR]}
else:
opts['renderer'] = mopts['renderer']
opts['failhard'] = mopts.get('failhard', False)
if mopts['state_top'].startswith('salt://'):
opts['state_top'] = mopts['state_top']
elif mopts['state_top'].startswith('/'):
opts['state_top'] = salt.utils.url.create(mopts['state_top'][1:])
else:
opts['state_top'] = salt.utils.url.create(mopts['state_top'])
opts['state_top_saltenv'] = mopts.get('state_top_saltenv', None)
opts['nodegroups'] = mopts.get('nodegroups', {})
opts['state_auto_order'] = mopts.get(
'state_auto_order',
opts['state_auto_order'])
opts['file_roots'] = mopts['file_roots']
opts['top_file_merging_strategy'] = mopts.get('top_file_merging_strategy',
opts.get('top_file_merging_strategy'))
opts['env_order'] = mopts.get('env_order', opts.get('env_order', []))
opts['default_top'] = mopts.get('default_top', opts.get('default_top'))
opts['state_events'] = mopts.get('state_events')
opts['state_aggregate'] = mopts.get('state_aggregate', opts.get('state_aggregate', False))
opts['jinja_lstrip_blocks'] = mopts.get('jinja_lstrip_blocks', False)
opts['jinja_trim_blocks'] = mopts.get('jinja_trim_blocks', False)
return opts def _get_envs(self):
'''
Pull the file server environments out of the master options
'''
envs = ['base']
if 'file_roots' in self.opts:
envs.extend(list(self.opts['file_roots']))
client_envs = self.client.envs()
env_order = self.opts.get('env_order', [])
client_envs = self.client.envs()
if env_order and client_envs:
client_env_list = self.client.envs()
env_intersection = set(env_order).intersection(client_env_list)
final_list = []
for ord_env in env_order:
if ord_env in env_intersection:
final_list.append(ord_env)
return set(final_list) elif env_order:
return set(env_order)
else:
for cenv in client_envs:
if cenv not in envs:
envs.append(cenv)
return set(envs) def get_tops(self):
'''
Gather the top files
'''
tops = DefaultOrderedDict(list)
include = DefaultOrderedDict(list)
done = DefaultOrderedDict(list)
found = 0 # did we find any contents in the top files?
# Gather initial top files
if self.opts['top_file_merging_strategy'] == 'same' and \
not self.opts['environment']:
if not self.opts['default_top']:
raise SaltRenderError('Top file merge strategy set to same, but no default_top '
'configuration option was set')
self.opts['environment'] = self.opts['default_top'] if self.opts['environment']:
contents = self.client.cache_file(
self.opts['state_top'],
self.opts['environment']
)
if contents:
found = 1
tops[self.opts['environment']] = [
compile_template(
contents,
self.state.rend,
self.state.opts['renderer'],
saltenv=self.opts['environment']
)
]
elif self.opts['top_file_merging_strategy'] == 'merge':
found = 0
if self.opts.get('state_top_saltenv', False):
saltenv = self.opts['state_top_saltenv']
contents = self.client.cache_file(
self.opts['state_top'],
saltenv
)
if contents:
found = found + 1
else:
log.debug('No contents loaded for env: {0}'.format(saltenv)) tops[saltenv].append(
compile_template(
contents,
self.state.rend,
self.state.opts['renderer'],
saltenv=saltenv
)
)
else:
for saltenv in self._get_envs():
contents = self.client.cache_file(
self.opts['state_top'],
saltenv
)
if contents:
found = found + 1
else:
log.debug('No contents loaded for env: {0}'.format(saltenv)) tops[saltenv].append(
compile_template(
contents,
self.state.rend,
self.state.opts['renderer'],
saltenv=saltenv
)
)
if found > 1:
log.warning('Top file merge strategy set to \'merge\' and multiple top files found. '
'Top file merging order is undefined; '
'for better results use \'same\' option') if found == 0:
log.error('No contents found in top file') # Search initial top files for includes
for saltenv, ctops in six.iteritems(tops):
for ctop in ctops:
if 'include' not in ctop:
continue
for sls in ctop['include']:
include[saltenv].append(sls)
ctop.pop('include')
# Go through the includes and pull out the extra tops and add them
while include:
pops = []
for saltenv, states in six.iteritems(include):
pops.append(saltenv)
if not states:
continue
for sls_match in states:
for sls in fnmatch.filter(self.avail[saltenv], sls_match):
if sls in done[saltenv]:
continue
tops[saltenv].append(
compile_template(
self.client.get_state(
sls,
saltenv
).get('dest', False),
self.state.rend,
self.state.opts['renderer'],
saltenv=saltenv
)
)
done[saltenv].append(sls)
for saltenv in pops:
if saltenv in include:
include.pop(saltenv)
return tops def merge_tops(self, tops):
'''
Cleanly merge the top files
'''
top = DefaultOrderedDict(OrderedDict)
for ctops in six.itervalues(tops):
for ctop in ctops:
for saltenv, targets in six.iteritems(ctop):
if saltenv == 'include':
continue
try:
for tgt in targets:
if tgt not in top[saltenv]:
top[saltenv][tgt] = ctop[saltenv][tgt]
continue
matches = []
states = set()
for comp in top[saltenv][tgt]:
if isinstance(comp, dict):
matches.append(comp)
if isinstance(comp, six.string_types):
states.add(comp)
top[saltenv][tgt] = matches
top[saltenv][tgt].extend(list(states))
except TypeError:
raise SaltRenderError('Unable to render top file. No targets found.')
return top def verify_tops(self, tops):
'''
Verify the contents of the top file data
'''
errors = []
if not isinstance(tops, dict):
errors.append('Top data was not formed as a dict')
# No further checks will work, bail out
return errors
for saltenv, matches in six.iteritems(tops):
if saltenv == 'include':
continue
if not isinstance(saltenv, six.string_types):
errors.append(
'Environment {0} in top file is not formed as a '
'string'.format(saltenv)
)
if saltenv == '':
errors.append('Empty saltenv statement in top file')
if not isinstance(matches, dict):
errors.append(
'The top file matches for saltenv {0} are not '
'formatted as a dict'.format(saltenv)
)
for slsmods in six.itervalues(matches):
if not isinstance(slsmods, list):
errors.append('Malformed topfile (state declarations not '
'formed as a list)')
continue
for slsmod in slsmods:
if isinstance(slsmod, dict):
# This value is a match option
for val in six.itervalues(slsmod):
if not val:
errors.append(
'Improperly formatted top file matcher '
'in saltenv {0}: {1} file'.format(
slsmod,
val
)
)
elif isinstance(slsmod, six.string_types):
# This is a sls module
if not slsmod:
errors.append(
'Environment {0} contains an empty sls '
'index'.format(saltenv)
) return errors def get_top(self):
'''
Returns the high data derived from the top file
'''
tops = self.get_tops()
return self.merge_tops(tops) def top_matches(self, top):
'''
Search through the top high data for matches and return the states
that this minion needs to execute. Returns:
{'saltenv': ['state1', 'state2', ...]}
'''
matches = {}
# pylint: disable=cell-var-from-loop
for saltenv, body in six.iteritems(top):
if self.opts['environment']:
if saltenv != self.opts['environment']:
continue
for match, data in six.iteritems(body):
def _filter_matches(_match, _data, _opts):
if isinstance(_data, six.string_types):
_data = [_data]
if self.matcher.confirm_top(
_match,
_data,
_opts
):
if saltenv not in matches:
matches[saltenv] = []
for item in _data:
if 'subfilter' in item:
_tmpdata = item.pop('subfilter')
for match, data in six.iteritems(_tmpdata):
_filter_matches(match, data, _opts)
if isinstance(item, six.string_types):
matches[saltenv].append(item)
_filter_matches(match, data, self.opts['nodegroups'])
ext_matches = self.client.ext_nodes()
for saltenv in ext_matches:
if saltenv in matches:
matches[saltenv] = list(
set(ext_matches[saltenv]).union(matches[saltenv]))
else:
matches[saltenv] = ext_matches[saltenv]
# pylint: enable=cell-var-from-loop
return matches def load_dynamic(self, matches):
'''
If autoload_dynamic_modules is True then automatically load the
dynamic modules
'''
if not self.opts['autoload_dynamic_modules']:
return
if self.opts.get('local', False):
syncd = self.state.functions['saltutil.sync_all'](list(matches),
refresh=False)
else:
syncd = self.state.functions['saltutil.sync_all'](list(matches),
refresh=False)
if syncd['grains']:
self.opts['grains'] = salt.loader.grains(self.opts)
self.state.opts['pillar'] = self.state._gather_pillar()
self.state.module_refresh() def render_state(self, sls, saltenv, mods, matches, local=False):
'''
Render a state file and retrieve all of the include states
'''
errors = []
if not local:
state_data = self.client.get_state(sls, saltenv)
fn_ = state_data.get('dest', False)
else:
fn_ = sls
if not os.path.isfile(fn_):
errors.append(
'Specified SLS {0} on local filesystem cannot '
'be found.'.format(sls)
)
if not fn_:
errors.append(
'Specified SLS {0} in saltenv {1} is not '
'available on the salt master or through a configured '
'fileserver'.format(sls, saltenv)
)
state = None
try:
state = compile_template(
fn_, self.state.rend, self.state.opts['renderer'], saltenv,
sls, rendered_sls=mods
)
except SaltRenderError as exc:
msg = 'Rendering SLS \'{0}:{1}\' failed: {2}'.format(
saltenv, sls, exc
)
log.critical(msg)
errors.append(msg)
except Exception as exc:
msg = 'Rendering SLS {0} failed, render error: {1}'.format(
sls, exc
)
log.critical(
msg,
# Show the traceback if the debug logging level is enabled
exc_info_on_loglevel=logging.DEBUG
)
errors.append('{0}\n{1}'.format(msg, traceback.format_exc()))
try:
mods.add('{0}:{1}'.format(saltenv, sls))
except AttributeError:
pass
if state:
if not isinstance(state, dict):
errors.append(
'SLS {0} does not render to a dictionary'.format(sls)
)
else:
include = []
if 'include' in state:
if not isinstance(state['include'], list):
err = ('Include Declaration in SLS {0} is not formed '
'as a list'.format(sls))
errors.append(err)
else:
include = state.pop('include') self._handle_extend(state, sls, saltenv, errors)
self._handle_exclude(state, sls, saltenv, errors)
self._handle_state_decls(state, sls, saltenv, errors) for inc_sls in include:
# inc_sls may take the form of:
# 'sls.to.include' <- same as {<saltenv>: 'sls.to.include'}
# {<env_key>: 'sls.to.include'}
# {'_xenv': 'sls.to.resolve'}
xenv_key = '_xenv' if isinstance(inc_sls, dict):
env_key, inc_sls = inc_sls.popitem()
else:
env_key = saltenv if env_key not in self.avail:
msg = ('Nonexistent saltenv {0!r} found in include '
'of {1!r} within SLS \'{2}:{3}\''
.format(env_key, inc_sls, saltenv, sls))
log.error(msg)
errors.append(msg)
continue if inc_sls.startswith('.'):
levels, include = \
re.match(r'^(\.+)(.*)$', inc_sls).groups()
level_count = len(levels)
p_comps = sls.split('.')
if state_data.get('source', '').endswith('/init.sls'):
p_comps.append('init')
if level_count > len(p_comps):
msg = ('Attempted relative include of {0!r} '
'within SLS \'{1}:{2}\' '
'goes beyond top level package '
.format(inc_sls, saltenv, sls))
log.error(msg)
errors.append(msg)
continue
inc_sls = '.'.join(p_comps[:-level_count] + [include]) if env_key != xenv_key:
if matches is None:
matches = []
# Resolve inc_sls in the specified environment
if env_key in matches or fnmatch.filter(self.avail[env_key], inc_sls):
resolved_envs = [env_key]
else:
resolved_envs = []
else:
# Resolve inc_sls in the subset of environment matches
resolved_envs = [
aenv for aenv in matches
if fnmatch.filter(self.avail[aenv], inc_sls)
] # An include must be resolved to a single environment, or
# the include must exist in the current environment
if len(resolved_envs) == 1 or saltenv in resolved_envs:
# Match inc_sls against the available states in the
# resolved env, matching wildcards in the process. If
# there were no matches, then leave inc_sls as the
# target so that the next recursion of render_state
# will recognize the error.
sls_targets = fnmatch.filter(
self.avail[saltenv],
inc_sls
) or [inc_sls] for sls_target in sls_targets:
r_env = resolved_envs[0] if len(resolved_envs) == 1 else saltenv
mod_tgt = '{0}:{1}'.format(r_env, sls_target)
if mod_tgt not in mods:
nstate, err = self.render_state(
sls_target,
r_env,
mods,
matches
)
if nstate:
self.merge_included_states(state, nstate, errors)
state.update(nstate)
if err:
errors.extend(err)
else:
msg = ''
if not resolved_envs:
msg = ('Unknown include: Specified SLS {0}: {1} is not available on the salt '
'master in saltenv(s): {2} '
).format(env_key,
inc_sls,
', '.join(matches) if env_key == xenv_key else env_key)
elif len(resolved_envs) > 1:
msg = ('Ambiguous include: Specified SLS {0}: {1} is available on the salt master '
'in multiple available saltenvs: {2}'
).format(env_key,
inc_sls,
', '.join(resolved_envs))
log.critical(msg)
errors.append(msg)
try:
self._handle_iorder(state)
except TypeError:
log.critical('Could not render SLS {0}. Syntax error detected.'.format(sls))
else:
state = {}
return state, errors def _handle_iorder(self, state):
'''
Take a state and apply the iorder system
'''
if self.opts['state_auto_order']:
for name in state:
for s_dec in state[name]:
if not isinstance(s_dec, six.string_types):
# PyDSL OrderedDict?
continue if not isinstance(state[name], dict):
# Include's or excludes as lists?
continue
if not isinstance(state[name][s_dec], list):
# Bad syntax, let the verify seq pick it up later on
continue found = False
if s_dec.startswith('_'):
continue for arg in state[name][s_dec]:
if isinstance(arg, dict):
if len(arg) > 0:
if next(six.iterkeys(arg)) == 'order':
found = True
if not found:
if not isinstance(state[name][s_dec], list):
# quite certainly a syntax error, managed elsewhere
continue
state[name][s_dec].append(
{'order': self.iorder}
)
self.iorder += 1
return state def _handle_state_decls(self, state, sls, saltenv, errors):
'''
Add sls and saltenv components to the state
'''
for name in state:
if not isinstance(state[name], dict):
if name == '__extend__':
continue
if name == '__exclude__':
continue if isinstance(state[name], six.string_types):
# Is this is a short state, it needs to be padded
if '.' in state[name]:
comps = state[name].split('.')
state[name] = {'__sls__': sls,
'__env__': saltenv,
comps[0]: [comps[1]]}
continue
errors.append(
'ID {0} in SLS {1} is not a dictionary'.format(name, sls)
)
continue
skeys = set()
for key in state[name]:
if key.startswith('_'):
continue
if not isinstance(state[name][key], list):
continue
if '.' in key:
comps = key.split('.')
# Salt doesn't support state files such as:
#
# /etc/redis/redis.conf:
# file.managed:
# - source: salt://redis/redis.conf
# - user: redis
# - group: redis
# - mode: 644
# file.comment:
# - regex: ^requirepass
if comps[0] in skeys:
errors.append(
'ID {0!r} in SLS {1!r} contains multiple state '
'declarations of the same type'.format(name, sls)
)
continue
state[name][comps[0]] = state[name].pop(key)
state[name][comps[0]].append(comps[1])
skeys.add(comps[0])
continue
skeys.add(key)
if '__sls__' not in state[name]:
state[name]['__sls__'] = sls
if '__env__' not in state[name]:
state[name]['__env__'] = saltenv def _handle_extend(self, state, sls, saltenv, errors):
'''
Take the extend dec out of state and apply to the highstate global
dec
'''
if 'extend' in state:
ext = state.pop('extend')
if not isinstance(ext, dict):
errors.append(('Extension value in SLS {0!r} is not a '
'dictionary').format(sls))
return
for name in ext:
if not isinstance(ext[name], dict):
errors.append(('Extension name {0!r} in SLS {1!r} is '
'not a dictionary'
.format(name, sls)))
continue
if '__sls__' not in ext[name]:
ext[name]['__sls__'] = sls
if '__env__' not in ext[name]:
ext[name]['__env__'] = saltenv
for key in ext[name]:
if key.startswith('_'):
continue
if not isinstance(ext[name][key], list):
continue
if '.' in key:
comps = key.split('.')
ext[name][comps[0]] = ext[name].pop(key)
ext[name][comps[0]].append(comps[1])
state.setdefault('__extend__', []).append(ext) def _handle_exclude(self, state, sls, saltenv, errors):
'''
Take the exclude dec out of the state and apply it to the highstate
global dec
'''
if 'exclude' in state:
exc = state.pop('exclude')
if not isinstance(exc, list):
err = ('Exclude Declaration in SLS {0} is not formed '
'as a list'.format(sls))
errors.append(err)
state.setdefault('__exclude__', []).extend(exc) def render_highstate(self, matches):
'''
Gather the state files and render them into a single unified salt
high data structure.
'''
highstate = self.building_highstate
all_errors = []
mods = set()
statefiles = []
for saltenv, states in six.iteritems(matches):
for sls_match in states:
try:
statefiles = fnmatch.filter(self.avail[saltenv], sls_match)
except KeyError:
all_errors.extend(['No matching salt environment for environment {0!r} found'.format(saltenv)])
# if we did not found any sls in the fileserver listing, this
# may be because the sls was generated or added later, we can
# try to directly execute it, and if it fails, anyway it will
# return the former error
if not statefiles:
statefiles = [sls_match] for sls in statefiles:
r_env = '{0}:{1}'.format(saltenv, sls)
if r_env in mods:
continue
state, errors = self.render_state(
sls, saltenv, mods, matches)
if state:
self.merge_included_states(highstate, state, errors)
for i, error in enumerate(errors[:]):
if 'is not available' in error:
# match SLS foobar in environment
this_sls = 'SLS {0} in saltenv'.format(
sls_match)
if this_sls in error:
errors[i] = (
'No matching sls found for {0!r} '
'in env {1!r}'.format(sls_match, saltenv))
all_errors.extend(errors) self.clean_duplicate_extends(highstate)
return highstate, all_errors def clean_duplicate_extends(self, highstate):
if '__extend__' in highstate:
highext = []
for items in (six.iteritems(ext) for ext in highstate['__extend__']):
for item in items:
if item not in highext:
highext.append(item)
highstate['__extend__'] = [{t[0]: t[1]} for t in highext] def merge_included_states(self, highstate, state, errors):
# The extend members can not be treated as globally unique:
if '__extend__' in state:
highstate.setdefault('__extend__',
[]).extend(state.pop('__extend__'))
if '__exclude__' in state:
highstate.setdefault('__exclude__',
[]).extend(state.pop('__exclude__'))
for id_ in state:
if id_ in highstate:
if highstate[id_] != state[id_]:
errors.append((
'Detected conflicting IDs, SLS'
' IDs need to be globally unique.\n The'
' conflicting ID is {0!r} and is found in SLS'
' \'{1}:{2}\' and SLS \'{3}:{4}\'').format(
id_,
highstate[id_]['__env__'],
highstate[id_]['__sls__'],
state[id_]['__env__'],
state[id_]['__sls__'])
)
try:
highstate.update(state)
except ValueError:
errors.append(
'Error when rendering state with contents: {0}'.format(state)
) def _check_pillar(self, force=False):
'''
Check the pillar for errors, refuse to run the state if there are
errors in the pillar and return the pillar errors
'''
if force:
return True
if '_errors' in self.state.opts['pillar']:
return False
return True def matches_whitelist(self, matches, whitelist):
'''
Reads over the matches and returns a matches dict with just the ones
that are in the whitelist
'''
if not whitelist:
return matches
ret_matches = {}
if not isinstance(whitelist, list):
whitelist = whitelist.split(',')
for env in matches:
for sls in matches[env]:
if sls in whitelist:
ret_matches[env] = ret_matches[env] if env in ret_matches else []
ret_matches[env].append(sls)
return ret_matches def call_highstate(self, exclude=None, cache=None, cache_name='highstate',
force=False, whitelist=None):
'''
Run the sequence to execute the salt highstate for this minion
'''
# Check that top file exists
tag_name = 'no_|-states_|-states_|-None'
ret = {tag_name: {
'result': False,
'comment': 'No states found for this minion',
'name': 'No States',
'changes': {},
'__run_num__': 0,
}}
cfn = os.path.join(
self.opts['cachedir'],
'{0}.cache.p'.format(cache_name)
) if cache:
if os.path.isfile(cfn):
with salt.utils.fopen(cfn, 'rb') as fp_:
high = self.serial.load(fp_)
return self.state.call_high(high)
# File exists so continue
err = []
try:
top = self.get_top()
except SaltRenderError as err:
ret[tag_name]['comment'] = 'Unable to render top file: '
ret[tag_name]['comment'] += err.error
return ret
except Exception:
trb = traceback.format_exc()
err.append(trb)
return err
err += self.verify_tops(top)
matches = self.top_matches(top)
if not matches:
msg = ('No Top file or external nodes data matches found')
ret[tag_name]['comment'] = msg
return ret
matches = self.matches_whitelist(matches, whitelist)
self.load_dynamic(matches)
if not self._check_pillar(force):
err += ['Pillar failed to render with the following messages:']
err += self.state.opts['pillar']['_errors']
else:
high, errors = self.render_highstate(matches)
if exclude:
if isinstance(exclude, str):
exclude = exclude.split(',')
if '__exclude__' in high:
high['__exclude__'].extend(exclude)
else:
high['__exclude__'] = exclude
err += errors
if err:
return err
if not high:
return ret
cumask = os.umask(0o77)
try:
if salt.utils.is_windows():
# Make sure cache file isn't read-only
self.state.functions['cmd.run']('attrib -R "{0}"'.format(cfn), output_loglevel='quiet')
with salt.utils.fopen(cfn, 'w+b') as fp_:
try:
self.serial.dump(high, fp_)
except TypeError:
# Can't serialize pydsl
pass
except (IOError, OSError):
msg = 'Unable to write to "state.highstate" cache file {0}'
log.error(msg.format(cfn)) os.umask(cumask)
return self.state.call_high(high) def compile_highstate(self):
'''
Return just the highstate or the errors
'''
err = []
top = self.get_top()
err += self.verify_tops(top)
matches = self.top_matches(top)
high, errors = self.render_highstate(matches)
err += errors if err:
return err return high def compile_low_chunks(self):
'''
Compile the highstate but don't run it, return the low chunks to
see exactly what the highstate will execute
'''
top = self.get_top()
matches = self.top_matches(top)
high, errors = self.render_highstate(matches) # If there is extension data reconcile it
high, ext_errors = self.state.reconcile_extend(high)
errors += ext_errors # Verify that the high data is structurally sound
errors += self.state.verify_high(high)
high, req_in_errors = self.state.requisite_in(high)
errors += req_in_errors
high = self.state.apply_exclude(high) if errors:
return errors # Compile and verify the raw chunks
chunks = self.state.compile_high_data(high) return chunks class HighState(BaseHighState):
'''
Generate and execute the salt "High State". The High State is the
compound state derived from a group of template files stored on the
salt master or in the local cache.
'''
# a stack of active HighState objects during a state.highstate run
stack = [] def __init__(self, opts, pillar=None, jid=None):
self.opts = opts
self.client = salt.fileclient.get_file_client(self.opts)
BaseHighState.__init__(self, opts)
self.state = State(self.opts, pillar, jid)
self.matcher = salt.minion.Matcher(self.opts) # tracks all pydsl state declarations globally across sls files
self._pydsl_all_decls = {} # a stack of current rendering Sls objects, maintained and used by the pydsl renderer.
self._pydsl_render_stack = [] def push_active(self):
self.stack.append(self) @classmethod
def clear_active(cls):
# Nuclear option
#
# Blow away the entire stack. Used primarily by the test runner but also
# useful in custom wrappers of the HighState class, to reset the stack
# to a fresh state.
cls.stack = [] @classmethod
def pop_active(cls):
cls.stack.pop() @classmethod
def get_active(cls):
try:
return cls.stack[-1]
except IndexError:
return None class MasterState(State):
'''
Create a State object for master side compiling
'''
def __init__(self, opts, minion):
State.__init__(self, opts) def load_modules(self, data=None):
'''
Load the modules into the state
'''
log.info('Loading fresh modules for state activity')
# Load a modified client interface that looks like the interface used
# from the minion, but uses remote execution
#
self.functions = salt.client.FunctionWrapper(
self.opts,
self.opts['id']
)
# Load the states, but they should not be used in this class apart
# from inspection
self.utils = salt.loader.utils(self.opts)
self.states = salt.loader.states(self.opts, self.functions, self.utils)
self.rend = salt.loader.render(self.opts, self.functions, states=self.states) class MasterHighState(HighState):
'''
Execute highstate compilation from the master
'''
def __init__(self, master_opts, minion_opts, grains, id_,
saltenv=None,
env=None):
if isinstance(env, six.string_types):
salt.utils.warn_until(
'Boron',
'Passing a salt environment should be done using \'saltenv\' '
'not \'env\'. This functionality will be removed in Salt '
'Boron.'
)
# Backwards compatibility
saltenv = env
# Force the fileclient to be local
opts = copy.deepcopy(minion_opts)
opts['file_client'] = 'local'
opts['file_roots'] = master_opts['master_roots']
opts['renderer'] = master_opts['renderer']
opts['state_top'] = master_opts['state_top']
opts['id'] = id_
opts['grains'] = grains
HighState.__init__(self, opts) class RemoteHighState(object):
'''
Manage gathering the data from the master
'''
def __init__(self, opts, grains):
self.opts = opts
self.grains = grains
self.serial = salt.payload.Serial(self.opts)
# self.auth = salt.crypt.SAuth(opts)
self.channel = salt.transport.Channel.factory(self.opts['master_uri']) def compile_master(self):
'''
Return the state data from the master
'''
load = {'grains': self.grains,
'opts': self.opts,
'cmd': '_master_state'}
try:
return self.channel.send(load, tries=3, timeout=72000)
except SaltReqTimeoutError:
return {}
saltstack对递归依赖条件(死循环依赖)的处理的更多相关文章
- TestNg依赖高级用法之强制依赖与顺序依赖------TestNg依赖详解(二)
TestNg使用dependsOnGroups属性来进行依赖测试, 测试方法依赖于某个或某些方法,这个/这些方法作为前置依赖条件 强制依赖:如果被依赖的某一个方法发生了异常,那么之后的方法都不会被执行 ...
- TestNg依靠先进的采用强制的依赖，并依赖序列的------TestNg依赖于特定的解释(两)
原创文章,版权所有所有,转载,归因:http://blog.csdn.net/wanghantong TestNg使用dependsOnGroups属性来进行依赖測试, 測试方法依赖于某个或某些方法, ...
- yii依赖注入和依赖注入容器
依赖注入和依赖注入容器¶ 为了降低代码耦合程度,提高项目的可维护性,Yii采用多许多当下最流行又相对成熟的设计模式,包括了依赖注入(Denpdency Injection, DI)和服务定位器(Ser ...
- yii2之依赖注入与依赖注入容器
一.为什么需要依赖注入 首先我们先不管什么是依赖注入,先来分析一下没有使用依赖注入会有什么样的结果.假设我们有一个gmail邮件服务类GMail,然后有另一个类User,User类需要使用发邮件的功能 ...
- gradle入门（1-8）gradle 的依赖查看、依赖排除和指定版本（需要验证！）
一.依赖查看 gradle dependencies 在gradle dependencies输出会有如下几种标记: 1.版本 : 唯一的依赖. 2.版本():还存在该库其他版本的依赖或者间接依赖,并 ...
- 【maven】maven查看项目依赖并解决依赖冲突的问题
一.问题 项目开发过程中,经常会遇到jar冲突,然后maven根据自己的规则进行冲突解决,导致项目在运行的过程中报错. 1.maven自动解决依赖冲突的规则是什么? 2.如何查看当前项目的maven的 ...
- BFS (1)算法模板 看是否需要分层 （2）拓扑排序——检测编译时的循环依赖 制定有依赖关系的任务的执行顺序 djkstra无非是将bfs模板中的deque修改为heapq
BFS模板,记住这5个: (1)针对树的BFS 1.1 无需分层遍历 from collections import deque def levelOrderTree(root): if not ro ...
- 大数据开发-从cogroup的实现来看join是宽依赖还是窄依赖
前面一篇文章提到大数据开发-Spark Join原理详解,本文从源码角度来看cogroup 的join实现 1.分析下面的代码 import org.apache.spark.rdd.RDD impo ...
- IoC模式（依赖、依赖倒置、依赖注入、控制反转）
1.依赖 依赖就是有联系,有地方使用到它就是有依赖它,一个系统不可能完全避免依赖.如果你的一个类或者模块在项目中没有用到它,恭喜你,可以从项目中剔除它或者排除它了,因为没有一个地方会依赖它.下面看一个 ...
随机推荐
- Android开发之InstanceState详解(转）
本文来自:http://www.cnblogs.com/hanyonglu/archive/2012/03/28/2420515.html 本文介绍Android中关于Activity的两个神秘方法: ...
- PLSQL developer连接不上64位Oracle的解决方法
PLSQL developer连接不上64位Oracle的解决方法 64位下装Oracle 11g 64位,PLSQL Developer使用出现问题. 问题描述: 登录对话框中,数据库下拉框为空: ...
- HTML5 canvas 中的线条样式
线条样式属性 lineCap 设置或返回线条的结束端点样式 butt 默认.向线条的每个末端添加平直的边缘. round 向线条的每个末端添加圆形线帽. ...
- C#中启动外部应用程序
C#中我们可以通过Process类直接启动外部应用程序 代码如下: Process p = new Process(); p.StartInfo.FileName ...
- mac 常用的开发工具
http://www.oschina.net/news/53946/mac-dev-tools 要清楚的认识到,我们寻找的不是开始按钮,而是程序入口,任何一个操作系统,用户要做的事情并不是找到开始菜单 ...
- Swift—类型检查与转换-备
继承会发生在子类和父类之间,是一系列类的继承关系. 例如:Person是类层次结构中的根类,Student是Person的直接子类,Worker是Person的直接子类. 这个继承关系类的具体实现代码 ...
- Web开发-各状态码的意思
常见的HTTP 1.1状态码以及它们对应的状态信息和含义 100 Continue 初始的请求已经接受,客户应当继续发送请求的其余部分.(HTTP 1.1新) 101 Switching Protoc ...
- Javascript OrderBy
要在js 实现orderBy基本知识就是 array.sortarray.sort(function(a,b){ a 表示 row 0 b 表示 row 1 它会loop多次你可以比较 if(a &g ...
- ossim/zabbix/logstash
http://dl528888.blog.51cto.com/2382721/1639579 http://307033.blog.51cto.com/297033/473666 http://lym ...
- FE: Responsive Web Design Overview
布局特点 1. 单列布局 vs 水平布局 手机屏幕宽度较小,因此多采用单列布局. 反观桌面网页设计,为了利用宽度,往往使用各种水平布局的组件,诸如水平导航栏,水平按钮组,水平分页等.然而水平布局 ...