deploy/cloud/deployment.py

   1 ###############################################################################
   2 # Copyright (c) 2015 Ericsson AB and others.
   3 # szilard.cserey@ericsson.com
   4 # All rights reserved. This program and the accompanying materials
   5 # are made available under the terms of the Apache License, Version 2.0
   6 # which accompanies this distribution, and is available at
   7 # http://www.apache.org/licenses/LICENSE-2.0
   8 ###############################################################################
   9
  10 import time
  11 import re
  12 import json
  13
  14 from common import (
  15     N,
  16     exec_cmd,
  17     parse,
  18     err,
  19     log,
  20 )
  21
  22 SEARCH_TEXT = '(err)'
  23 LOG_FILE = '/var/log/puppet.log'
  24 GREP_LINES_OF_LEADING_CONTEXT = 100
  25 GREP_LINES_OF_TRAILING_CONTEXT = 100
  26 LIST_OF_CHAR_TO_BE_ESCAPED = ['[', ']', '"']
  27 ERROR_MSGS = ['Critical nodes are not available for deployment',
  28               'offline. Remove them from environment and try again.']
  29
  30
  31 class DeployNotStart(Exception):
  32     """Unable to start deployment"""
  33
  34
  35 class NodesGoOffline(Exception):
  36     """Nodes goes offline during deployment"""
  37
  38
  39 class Deployment(object):
  40
  41     def __init__(self, dea, yaml_config_dir, env_id, node_id_roles_dict,
  42                  no_health_check, deploy_timeout):
  43         self.dea = dea
  44         self.yaml_config_dir = yaml_config_dir
  45         self.env_id = env_id
  46         self.node_id_roles_dict = node_id_roles_dict
  47         self.no_health_check = no_health_check
  48         self.deploy_timeout = deploy_timeout
  49         self.pattern = re.compile(
  50             '\d\d\d\d-\d\d-\d\d\s\d\d:\d\d:\d\d')
  51
  52     def collect_error_logs(self):
  53         for node_id, roles_blade in self.node_id_roles_dict.iteritems():
  54             log_list = []
  55             cmd = ('ssh -q node-%s grep \'"%s"\' %s'
  56                    % (node_id, SEARCH_TEXT, LOG_FILE))
  57             results, _ = exec_cmd(cmd, False)
  58             for result in results.splitlines():
  59                 log_msg = ''
  60
  61                 sub_cmd = '"%s" %s' % (result, LOG_FILE)
  62                 for c in LIST_OF_CHAR_TO_BE_ESCAPED:
  63                     sub_cmd = sub_cmd.replace(c, '\%s' % c)
  64                 grep_cmd = ('grep -B%s %s'
  65                             % (GREP_LINES_OF_LEADING_CONTEXT, sub_cmd))
  66                 cmd = ('ssh -q node-%s "%s"' % (node_id, grep_cmd))
  67
  68                 details, _ = exec_cmd(cmd, False)
  69                 details_list = details.splitlines()
  70
  71                 found_prev_log = False
  72                 for i in range(len(details_list) - 2, -1, -1):
  73                     if self.pattern.match(details_list[i]):
  74                         found_prev_log = True
  75                         break
  76                 if found_prev_log:
  77                     log_msg += '\n'.join(details_list[i:-1]) + '\n'
  78
  79                 grep_cmd = ('grep -A%s %s'
  80                             % (GREP_LINES_OF_TRAILING_CONTEXT, sub_cmd))
  81                 cmd = ('ssh -q node-%s "%s"' % (node_id, grep_cmd))
  82
  83                 details, _ = exec_cmd(cmd, False)
  84                 details_list = details.splitlines()
  85
  86                 found_next_log = False
  87                 for i in range(1, len(details_list)):
  88                     if self.pattern.match(details_list[i]):
  89                         found_next_log = True
  90                         break
  91                 if found_next_log:
  92                     log_msg += '\n'.join(details_list[:i])
  93                 else:
  94                     log_msg += details
  95
  96                 if log_msg:
  97                     log_list.append(log_msg)
  98
  99             if log_list:
 100                 role = ('controller' if 'controller' in roles_blade[0]
 101                         else 'compute host')
 102                 log('_' * 40 + 'Errors in node-%s %s' % (node_id, role)
 103                     + '_' * 40)
 104                 for log_msg in log_list:
 105                     print(log_msg + '\n')
 106
 107     def run_deploy(self):
 108         SLEEP_TIME = 60
 109         abort_after = 60 * int(self.deploy_timeout)
 110         start = time.time()
 111
 112         log('Starting deployment of environment %s' % self.env_id)
 113         deploy_id = None
 114         ready = False
 115         timeout = False
 116
 117         attempts = 5
 118         while attempts > 0:
 119             try:
 120                 if time.time() > start + abort_after:
 121                     timeout = True
 122                     break
 123                 if not deploy_id:
 124                     deploy_id = self._start_deploy_task()
 125                 sts, prg, msg = self._deployment_status(deploy_id)
 126                 if sts == 'error':
 127                     log('Error during deployment: {}'.format(msg))
 128                     break
 129                 if sts == 'running':
 130                     log('Environment deployment progress: {}%'.format(prg))
 131                 elif sts == 'ready':
 132                     ready = True
 133                     break
 134                 time.sleep(SLEEP_TIME)
 135             except (DeployNotStart, NodesGoOffline) as e:
 136                 log(e)
 137                 attempts -= 1
 138                 deploy_id = None
 139                 time.sleep(SLEEP_TIME * attempts)
 140
 141         if timeout:
 142             err('Deployment timed out, environment %s is not operational, '
 143                 'snapshot will not be performed'
 144                 % self.env_id)
 145         if ready:
 146             log('Environment %s successfully deployed'
 147                 % self.env_id)
 148         else:
 149             self.collect_error_logs()
 150             err('Deployment failed, environment %s is not operational'
 151                 % self.env_id, self.collect_logs)
 152
 153     def _start_deploy_task(self):
 154         out, _ = exec_cmd('fuel2 env deploy {}'.format(self.env_id), False)
 155         id = self._deployment_task_id(out)
 156         return id
 157
 158     def _deployment_task_id(self, response):
 159         response = str(response)
 160         if response.startswith('Deployment task with id'):
 161             for s in response.split():
 162                 if s.isdigit():
 163                     return int(s)
 164         raise DeployNotStart('Unable to start deployment: {}'.format(response))
 165
 166     def _deployment_status(self, id):
 167         task = self._task_fields(id)
 168         if task['status'] == 'error':
 169             if any(msg in task['message'] for msg in ERROR_MSGS):
 170                 raise NodesGoOffline(task['message'])
 171         return task['status'], task['progress'], task['message']
 172
 173     def _task_fields(self, id):
 174         try:
 175             out, _ = exec_cmd('fuel2 task show {} -f json'.format(id), False)
 176             task_info = json.loads(out)
 177             properties = {}
 178             # for 9.0 this can be list of dicts or dict
 179             # see https://bugs.launchpad.net/fuel/+bug/1625518
 180             if isinstance(task_info, list):
 181                 for d in task_info:
 182                         properties.update({d['Field']: d['Value']})
 183             else:
 184                 return task_info
 185             return properties
 186         except ValueError as e:
 187             err('Unable to fetch task info: {}'.format(e))
 188
 189     def collect_logs(self):
 190         log('Cleaning out any previous deployment logs')
 191         exec_cmd('rm -f /var/log/remote/fuel-snapshot-*', False)
 192         exec_cmd('rm -f /root/deploy-*', False)
 193         log('Generating Fuel deploy snap-shot')
 194         if exec_cmd('fuel snapshot < /dev/null &> snapshot.log', False)[1] != 0:
 195             log('Could not create a Fuel snapshot')
 196         else:
 197             exec_cmd('mv /root/fuel-snapshot* /var/log/remote/', False)
 198
 199         log('Collecting all Fuel Snapshot & deploy log files')
 200         r, _ = exec_cmd('tar -czhf /root/deploy-%s.log.tar.gz /var/log/remote' % time.strftime("%Y%m%d-%H%M%S"), False)
 201         log(r)
 202
 203     def verify_node_status(self):
 204         node_list = parse(exec_cmd('fuel --env %s node' % self.env_id))
 205         failed_nodes = []
 206         for node in node_list:
 207             if node[N['status']] != 'ready':
 208                 failed_nodes.append((node[N['id']], node[N['status']]))
 209
 210         if failed_nodes:
 211             summary = ''
 212             for node, status in failed_nodes:
 213                 summary += '[node %s, status %s]\n' % (node, status)
 214             err('Deployment failed: %s' % summary, self.collect_logs)
 215
 216     def health_check(self):
 217         log('Now running sanity and smoke health checks')
 218         r = exec_cmd('fuel health --env %s --check sanity,smoke --force' % self.env_id)
 219         log(r)
 220         if 'failure' in r:
 221             err('Healthcheck failed!', self.collect_logs)
 222
 223     def deploy(self):
 224         self.run_deploy()
 225         self.verify_node_status()
 226         if not self.no_health_check:
 227             self.health_check()
 228         self.collect_logs()