root / lib / bootstrap.py @ 95a74ef3
History | View | Annotate | Download (33.3 kB)
1 | a0c9f010 | Michael Hanselmann | #
|
---|---|---|---|
2 | a0c9f010 | Michael Hanselmann | #
|
3 | a0c9f010 | Michael Hanselmann | |
4 | 57dc299a | Iustin Pop | # Copyright (C) 2006, 2007, 2008, 2010, 2011, 2012 Google Inc.
|
5 | a0c9f010 | Michael Hanselmann | #
|
6 | a0c9f010 | Michael Hanselmann | # This program is free software; you can redistribute it and/or modify
|
7 | a0c9f010 | Michael Hanselmann | # it under the terms of the GNU General Public License as published by
|
8 | a0c9f010 | Michael Hanselmann | # the Free Software Foundation; either version 2 of the License, or
|
9 | a0c9f010 | Michael Hanselmann | # (at your option) any later version.
|
10 | a0c9f010 | Michael Hanselmann | #
|
11 | a0c9f010 | Michael Hanselmann | # This program is distributed in the hope that it will be useful, but
|
12 | a0c9f010 | Michael Hanselmann | # WITHOUT ANY WARRANTY; without even the implied warranty of
|
13 | a0c9f010 | Michael Hanselmann | # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
14 | a0c9f010 | Michael Hanselmann | # General Public License for more details.
|
15 | a0c9f010 | Michael Hanselmann | #
|
16 | a0c9f010 | Michael Hanselmann | # You should have received a copy of the GNU General Public License
|
17 | a0c9f010 | Michael Hanselmann | # along with this program; if not, write to the Free Software
|
18 | a0c9f010 | Michael Hanselmann | # Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA
|
19 | a0c9f010 | Michael Hanselmann | # 02110-1301, USA.
|
20 | a0c9f010 | Michael Hanselmann | |
21 | a0c9f010 | Michael Hanselmann | |
22 | a0c9f010 | Michael Hanselmann | """Functions to bootstrap a new cluster.
|
23 | a0c9f010 | Michael Hanselmann |
|
24 | a0c9f010 | Michael Hanselmann | """
|
25 | a0c9f010 | Michael Hanselmann | |
26 | a0c9f010 | Michael Hanselmann | import os |
27 | a0c9f010 | Michael Hanselmann | import os.path |
28 | a0c9f010 | Michael Hanselmann | import re |
29 | b1b6ea87 | Iustin Pop | import logging |
30 | d693c864 | Iustin Pop | import time |
31 | a0c9f010 | Michael Hanselmann | |
32 | a0c9f010 | Michael Hanselmann | from ganeti import rpc |
33 | a0c9f010 | Michael Hanselmann | from ganeti import ssh |
34 | a0c9f010 | Michael Hanselmann | from ganeti import utils |
35 | a0c9f010 | Michael Hanselmann | from ganeti import errors |
36 | a0c9f010 | Michael Hanselmann | from ganeti import config |
37 | a0c9f010 | Michael Hanselmann | from ganeti import constants |
38 | b9eeeb02 | Michael Hanselmann | from ganeti import objects |
39 | a0c9f010 | Michael Hanselmann | from ganeti import ssconf |
40 | a33848a5 | Guido Trotter | from ganeti import serializer |
41 | a5728081 | Guido Trotter | from ganeti import hypervisor |
42 | a721e23a | Luca Bigliardi | from ganeti import bdev |
43 | a744b676 | Manuel Franceschini | from ganeti import netutils |
44 | d367b66c | Manuel Franceschini | from ganeti import backend |
45 | 3b6b6129 | Michael Hanselmann | from ganeti import luxi |
46 | ff699aa9 | Michael Hanselmann | from ganeti import jstore |
47 | 3b6b6129 | Michael Hanselmann | |
48 | a0c9f010 | Michael Hanselmann | |
49 | 88b92fe3 | Guido Trotter | # ec_id for InitConfig's temporary reservation manager
|
50 | 88b92fe3 | Guido Trotter | _INITCONF_ECID = "initconfig-ecid"
|
51 | 88b92fe3 | Guido Trotter | |
52 | 3b6b6129 | Michael Hanselmann | #: After how many seconds daemon must be responsive
|
53 | 3b6b6129 | Michael Hanselmann | _DAEMON_READY_TIMEOUT = 10.0
|
54 | 3b6b6129 | Michael Hanselmann | |
55 | e38220e4 | Michael Hanselmann | |
56 | 531baf8e | Iustin Pop | def _InitSSHSetup(): |
57 | a0c9f010 | Michael Hanselmann | """Setup the SSH configuration for the cluster.
|
58 | a0c9f010 | Michael Hanselmann |
|
59 | a0c9f010 | Michael Hanselmann | This generates a dsa keypair for root, adds the pub key to the
|
60 | a0c9f010 | Michael Hanselmann | permitted hosts and adds the hostkey to its own known hosts.
|
61 | a0c9f010 | Michael Hanselmann |
|
62 | a0c9f010 | Michael Hanselmann | """
|
63 | a0c9f010 | Michael Hanselmann | priv_key, pub_key, auth_keys = ssh.GetUserFiles(constants.GANETI_RUNAS) |
64 | a0c9f010 | Michael Hanselmann | |
65 | a0c9f010 | Michael Hanselmann | for name in priv_key, pub_key: |
66 | a0c9f010 | Michael Hanselmann | if os.path.exists(name):
|
67 | a0c9f010 | Michael Hanselmann | utils.CreateBackup(name) |
68 | a0c9f010 | Michael Hanselmann | utils.RemoveFile(name) |
69 | a0c9f010 | Michael Hanselmann | |
70 | a0c9f010 | Michael Hanselmann | result = utils.RunCmd(["ssh-keygen", "-t", "dsa", |
71 | a0c9f010 | Michael Hanselmann | "-f", priv_key,
|
72 | a0c9f010 | Michael Hanselmann | "-q", "-N", ""]) |
73 | a0c9f010 | Michael Hanselmann | if result.failed:
|
74 | a0c9f010 | Michael Hanselmann | raise errors.OpExecError("Could not generate ssh keypair, error %s" % |
75 | a0c9f010 | Michael Hanselmann | result.output) |
76 | a0c9f010 | Michael Hanselmann | |
77 | 7a0156dc | Luca Bigliardi | utils.AddAuthorizedKey(auth_keys, utils.ReadFile(pub_key)) |
78 | a0c9f010 | Michael Hanselmann | |
79 | a0c9f010 | Michael Hanselmann | |
80 | c008906b | Michael Hanselmann | def GenerateHmacKey(file_name): |
81 | c008906b | Michael Hanselmann | """Writes a new HMAC key.
|
82 | c008906b | Michael Hanselmann |
|
83 | c008906b | Michael Hanselmann | @type file_name: str
|
84 | c008906b | Michael Hanselmann | @param file_name: Path to output file
|
85 | c008906b | Michael Hanselmann |
|
86 | c008906b | Michael Hanselmann | """
|
87 | 43575108 | Michael Hanselmann | utils.WriteFile(file_name, data="%s\n" % utils.GenerateSecret(), mode=0400, |
88 | 43575108 | Michael Hanselmann | backup=True)
|
89 | 43575108 | Michael Hanselmann | |
90 | 43575108 | Michael Hanselmann | |
91 | b6267745 | Andrea Spadaccini | def GenerateClusterCrypto(new_cluster_cert, new_rapi_cert, new_spice_cert, |
92 | b6267745 | Andrea Spadaccini | new_confd_hmac_key, new_cds, |
93 | b6267745 | Andrea Spadaccini | rapi_cert_pem=None, spice_cert_pem=None, |
94 | b6267745 | Andrea Spadaccini | spice_cacert_pem=None, cds=None, |
95 | aeefe835 | Iustin Pop | nodecert_file=constants.NODED_CERT_FILE, |
96 | aeefe835 | Iustin Pop | rapicert_file=constants.RAPI_CERT_FILE, |
97 | b6267745 | Andrea Spadaccini | spicecert_file=constants.SPICE_CERT_FILE, |
98 | b6267745 | Andrea Spadaccini | spicecacert_file=constants.SPICE_CACERT_FILE, |
99 | fc0726b9 | Michael Hanselmann | hmackey_file=constants.CONFD_HMAC_KEY, |
100 | fc0726b9 | Michael Hanselmann | cds_file=constants.CLUSTER_DOMAIN_SECRET_FILE): |
101 | 43575108 | Michael Hanselmann | """Updates the cluster certificates, keys and secrets.
|
102 | 43575108 | Michael Hanselmann |
|
103 | 43575108 | Michael Hanselmann | @type new_cluster_cert: bool
|
104 | 43575108 | Michael Hanselmann | @param new_cluster_cert: Whether to generate a new cluster certificate
|
105 | 43575108 | Michael Hanselmann | @type new_rapi_cert: bool
|
106 | 43575108 | Michael Hanselmann | @param new_rapi_cert: Whether to generate a new RAPI certificate
|
107 | b6267745 | Andrea Spadaccini | @type new_spice_cert: bool
|
108 | b6267745 | Andrea Spadaccini | @param new_spice_cert: Whether to generate a new SPICE certificate
|
109 | 6b7d5878 | Michael Hanselmann | @type new_confd_hmac_key: bool
|
110 | 6b7d5878 | Michael Hanselmann | @param new_confd_hmac_key: Whether to generate a new HMAC key
|
111 | 3db3eb2a | Michael Hanselmann | @type new_cds: bool
|
112 | 3db3eb2a | Michael Hanselmann | @param new_cds: Whether to generate a new cluster domain secret
|
113 | 43575108 | Michael Hanselmann | @type rapi_cert_pem: string
|
114 | 43575108 | Michael Hanselmann | @param rapi_cert_pem: New RAPI certificate in PEM format
|
115 | b6267745 | Andrea Spadaccini | @type spice_cert_pem: string
|
116 | b6267745 | Andrea Spadaccini | @param spice_cert_pem: New SPICE certificate in PEM format
|
117 | b6267745 | Andrea Spadaccini | @type spice_cacert_pem: string
|
118 | b6267745 | Andrea Spadaccini | @param spice_cacert_pem: Certificate of the CA that signed the SPICE
|
119 | b6267745 | Andrea Spadaccini | certificate, in PEM format
|
120 | 3db3eb2a | Michael Hanselmann | @type cds: string
|
121 | 3db3eb2a | Michael Hanselmann | @param cds: New cluster domain secret
|
122 | aeefe835 | Iustin Pop | @type nodecert_file: string
|
123 | aeefe835 | Iustin Pop | @param nodecert_file: optional override of the node cert file path
|
124 | aeefe835 | Iustin Pop | @type rapicert_file: string
|
125 | aeefe835 | Iustin Pop | @param rapicert_file: optional override of the rapi cert file path
|
126 | b6267745 | Andrea Spadaccini | @type spicecert_file: string
|
127 | b6267745 | Andrea Spadaccini | @param spicecert_file: optional override of the spice cert file path
|
128 | b6267745 | Andrea Spadaccini | @type spicecacert_file: string
|
129 | b6267745 | Andrea Spadaccini | @param spicecacert_file: optional override of the spice CA cert file path
|
130 | aeefe835 | Iustin Pop | @type hmackey_file: string
|
131 | aeefe835 | Iustin Pop | @param hmackey_file: optional override of the hmac key file path
|
132 | 43575108 | Michael Hanselmann |
|
133 | 43575108 | Michael Hanselmann | """
|
134 | 168c1de2 | Michael Hanselmann | # noded SSL certificate
|
135 | aeefe835 | Iustin Pop | cluster_cert_exists = os.path.exists(nodecert_file) |
136 | 43575108 | Michael Hanselmann | if new_cluster_cert or not cluster_cert_exists: |
137 | 43575108 | Michael Hanselmann | if cluster_cert_exists:
|
138 | aeefe835 | Iustin Pop | utils.CreateBackup(nodecert_file) |
139 | 43575108 | Michael Hanselmann | |
140 | aeefe835 | Iustin Pop | logging.debug("Generating new cluster certificate at %s", nodecert_file)
|
141 | af2ae1c0 | Iustin Pop | utils.GenerateSelfSignedSslCert(nodecert_file) |
142 | 43575108 | Michael Hanselmann | |
143 | 6b7d5878 | Michael Hanselmann | # confd HMAC key
|
144 | aeefe835 | Iustin Pop | if new_confd_hmac_key or not os.path.exists(hmackey_file): |
145 | aeefe835 | Iustin Pop | logging.debug("Writing new confd HMAC key to %s", hmackey_file)
|
146 | aeefe835 | Iustin Pop | GenerateHmacKey(hmackey_file) |
147 | 43575108 | Michael Hanselmann | |
148 | 43575108 | Michael Hanselmann | # RAPI
|
149 | aeefe835 | Iustin Pop | rapi_cert_exists = os.path.exists(rapicert_file) |
150 | 43575108 | Michael Hanselmann | |
151 | 43575108 | Michael Hanselmann | if rapi_cert_pem:
|
152 | 43575108 | Michael Hanselmann | # Assume rapi_pem contains a valid PEM-formatted certificate and key
|
153 | aeefe835 | Iustin Pop | logging.debug("Writing RAPI certificate at %s", rapicert_file)
|
154 | aeefe835 | Iustin Pop | utils.WriteFile(rapicert_file, data=rapi_cert_pem, backup=True)
|
155 | 43575108 | Michael Hanselmann | |
156 | 43575108 | Michael Hanselmann | elif new_rapi_cert or not rapi_cert_exists: |
157 | 43575108 | Michael Hanselmann | if rapi_cert_exists:
|
158 | aeefe835 | Iustin Pop | utils.CreateBackup(rapicert_file) |
159 | 43575108 | Michael Hanselmann | |
160 | aeefe835 | Iustin Pop | logging.debug("Generating new RAPI certificate at %s", rapicert_file)
|
161 | af2ae1c0 | Iustin Pop | utils.GenerateSelfSignedSslCert(rapicert_file) |
162 | c008906b | Michael Hanselmann | |
163 | b6267745 | Andrea Spadaccini | # SPICE
|
164 | b6267745 | Andrea Spadaccini | spice_cert_exists = os.path.exists(spicecert_file) |
165 | b6267745 | Andrea Spadaccini | spice_cacert_exists = os.path.exists(spicecacert_file) |
166 | b6267745 | Andrea Spadaccini | if spice_cert_pem:
|
167 | b6267745 | Andrea Spadaccini | # spice_cert_pem implies also spice_cacert_pem
|
168 | b6267745 | Andrea Spadaccini | logging.debug("Writing SPICE certificate at %s", spicecert_file)
|
169 | b6267745 | Andrea Spadaccini | utils.WriteFile(spicecert_file, data=spice_cert_pem, backup=True)
|
170 | b6267745 | Andrea Spadaccini | logging.debug("Writing SPICE CA certificate at %s", spicecacert_file)
|
171 | b6267745 | Andrea Spadaccini | utils.WriteFile(spicecacert_file, data=spice_cacert_pem, backup=True)
|
172 | b6267745 | Andrea Spadaccini | elif new_spice_cert or not spice_cert_exists: |
173 | b6267745 | Andrea Spadaccini | if spice_cert_exists:
|
174 | b6267745 | Andrea Spadaccini | utils.CreateBackup(spicecert_file) |
175 | b6267745 | Andrea Spadaccini | if spice_cacert_exists:
|
176 | b6267745 | Andrea Spadaccini | utils.CreateBackup(spicecacert_file) |
177 | b6267745 | Andrea Spadaccini | |
178 | b6267745 | Andrea Spadaccini | logging.debug("Generating new self-signed SPICE certificate at %s",
|
179 | b6267745 | Andrea Spadaccini | spicecert_file) |
180 | b6267745 | Andrea Spadaccini | (_, cert_pem) = utils.GenerateSelfSignedSslCert(spicecert_file) |
181 | b6267745 | Andrea Spadaccini | |
182 | b6267745 | Andrea Spadaccini | # Self-signed certificate -> the public certificate is also the CA public
|
183 | b6267745 | Andrea Spadaccini | # certificate
|
184 | b6267745 | Andrea Spadaccini | logging.debug("Writing the public certificate to %s",
|
185 | b6267745 | Andrea Spadaccini | spicecert_file) |
186 | b6267745 | Andrea Spadaccini | utils.io.WriteFile(spicecacert_file, mode=0400, data=cert_pem)
|
187 | b6267745 | Andrea Spadaccini | |
188 | 3db3eb2a | Michael Hanselmann | # Cluster domain secret
|
189 | 3db3eb2a | Michael Hanselmann | if cds:
|
190 | fc0726b9 | Michael Hanselmann | logging.debug("Writing cluster domain secret to %s", cds_file)
|
191 | fc0726b9 | Michael Hanselmann | utils.WriteFile(cds_file, data=cds, backup=True)
|
192 | fc0726b9 | Michael Hanselmann | |
193 | fc0726b9 | Michael Hanselmann | elif new_cds or not os.path.exists(cds_file): |
194 | fc0726b9 | Michael Hanselmann | logging.debug("Generating new cluster domain secret at %s", cds_file)
|
195 | fc0726b9 | Michael Hanselmann | GenerateHmacKey(cds_file) |
196 | 3db3eb2a | Michael Hanselmann | |
197 | c008906b | Michael Hanselmann | |
198 | 8f215968 | Michael Hanselmann | def _InitGanetiServerSetup(master_name): |
199 | 40a97d80 | Michael Hanselmann | """Setup the necessary configuration for the initial node daemon.
|
200 | 40a97d80 | Michael Hanselmann |
|
201 | 40a97d80 | Michael Hanselmann | This creates the nodepass file containing the shared password for
|
202 | 600535f0 | Manuel Franceschini | the cluster, generates the SSL certificate and starts the node daemon.
|
203 | 600535f0 | Manuel Franceschini |
|
204 | 600535f0 | Manuel Franceschini | @type master_name: str
|
205 | 600535f0 | Manuel Franceschini | @param master_name: Name of the master node
|
206 | 40a97d80 | Michael Hanselmann |
|
207 | 40a97d80 | Michael Hanselmann | """
|
208 | 43575108 | Michael Hanselmann | # Generate cluster secrets
|
209 | b6267745 | Andrea Spadaccini | GenerateClusterCrypto(True, False, False, False, False) |
210 | 4a34c5cf | Guido Trotter | |
211 | f154a7a3 | Michael Hanselmann | result = utils.RunCmd([constants.DAEMON_UTIL, "start", constants.NODED])
|
212 | a0c9f010 | Michael Hanselmann | if result.failed:
|
213 | a0c9f010 | Michael Hanselmann | raise errors.OpExecError("Could not start the node daemon, command %s" |
214 | a0c9f010 | Michael Hanselmann | " had exitcode %s and error %s" %
|
215 | a0c9f010 | Michael Hanselmann | (result.cmd, result.exit_code, result.output)) |
216 | a0c9f010 | Michael Hanselmann | |
217 | 5627f375 | Michael Hanselmann | _WaitForNodeDaemon(master_name) |
218 | 5627f375 | Michael Hanselmann | |
219 | 5627f375 | Michael Hanselmann | |
220 | 5627f375 | Michael Hanselmann | def _WaitForNodeDaemon(node_name): |
221 | 5627f375 | Michael Hanselmann | """Wait for node daemon to become responsive.
|
222 | 5627f375 | Michael Hanselmann |
|
223 | 5627f375 | Michael Hanselmann | """
|
224 | d3833ebd | Michael Hanselmann | def _CheckNodeDaemon(): |
225 | bd6d1202 | Renรฉ Nussbaumer | # Pylint bug <http://www.logilab.org/ticket/35642>
|
226 | bd6d1202 | Renรฉ Nussbaumer | # pylint: disable=E1101
|
227 | db04ce5d | Michael Hanselmann | result = rpc.BootstrapRunner().call_version([node_name])[node_name] |
228 | d3833ebd | Michael Hanselmann | if result.fail_msg:
|
229 | d3833ebd | Michael Hanselmann | raise utils.RetryAgain()
|
230 | 8f215968 | Michael Hanselmann | |
231 | d3833ebd | Michael Hanselmann | try:
|
232 | 3b6b6129 | Michael Hanselmann | utils.Retry(_CheckNodeDaemon, 1.0, _DAEMON_READY_TIMEOUT)
|
233 | d3833ebd | Michael Hanselmann | except utils.RetryTimeout:
|
234 | 5627f375 | Michael Hanselmann | raise errors.OpExecError("Node daemon on %s didn't answer queries within" |
235 | 3b6b6129 | Michael Hanselmann | " %s seconds" % (node_name, _DAEMON_READY_TIMEOUT))
|
236 | 3b6b6129 | Michael Hanselmann | |
237 | 3b6b6129 | Michael Hanselmann | |
238 | 3b6b6129 | Michael Hanselmann | def _WaitForMasterDaemon(): |
239 | 3b6b6129 | Michael Hanselmann | """Wait for master daemon to become responsive.
|
240 | 3b6b6129 | Michael Hanselmann |
|
241 | 3b6b6129 | Michael Hanselmann | """
|
242 | 3b6b6129 | Michael Hanselmann | def _CheckMasterDaemon(): |
243 | 3b6b6129 | Michael Hanselmann | try:
|
244 | 3b6b6129 | Michael Hanselmann | cl = luxi.Client() |
245 | 3b6b6129 | Michael Hanselmann | (cluster_name, ) = cl.QueryConfigValues(["cluster_name"])
|
246 | 3b6b6129 | Michael Hanselmann | except Exception: |
247 | 3b6b6129 | Michael Hanselmann | raise utils.RetryAgain()
|
248 | 3b6b6129 | Michael Hanselmann | |
249 | 3b6b6129 | Michael Hanselmann | logging.debug("Received cluster name %s from master", cluster_name)
|
250 | 3b6b6129 | Michael Hanselmann | |
251 | 3b6b6129 | Michael Hanselmann | try:
|
252 | 3b6b6129 | Michael Hanselmann | utils.Retry(_CheckMasterDaemon, 1.0, _DAEMON_READY_TIMEOUT)
|
253 | 3b6b6129 | Michael Hanselmann | except utils.RetryTimeout:
|
254 | 3b6b6129 | Michael Hanselmann | raise errors.OpExecError("Master daemon didn't answer queries within" |
255 | 3b6b6129 | Michael Hanselmann | " %s seconds" % _DAEMON_READY_TIMEOUT)
|
256 | 5627f375 | Michael Hanselmann | |
257 | a0c9f010 | Michael Hanselmann | |
258 | 0e3baaf3 | Iustin Pop | def _InitFileStorage(file_storage_dir): |
259 | 0e3baaf3 | Iustin Pop | """Initialize if needed the file storage.
|
260 | 0e3baaf3 | Iustin Pop |
|
261 | 0e3baaf3 | Iustin Pop | @param file_storage_dir: the user-supplied value
|
262 | 0e3baaf3 | Iustin Pop | @return: either empty string (if file storage was disabled at build
|
263 | 0e3baaf3 | Iustin Pop | time) or the normalized path to the storage directory
|
264 | 0e3baaf3 | Iustin Pop |
|
265 | 0e3baaf3 | Iustin Pop | """
|
266 | 0e3baaf3 | Iustin Pop | file_storage_dir = os.path.normpath(file_storage_dir) |
267 | 0e3baaf3 | Iustin Pop | |
268 | 0e3baaf3 | Iustin Pop | if not os.path.isabs(file_storage_dir): |
269 | 0376655e | Guido Trotter | raise errors.OpPrereqError("File storage directory '%s' is not an absolute" |
270 | 0376655e | Guido Trotter | " path" % file_storage_dir, errors.ECODE_INVAL)
|
271 | 0e3baaf3 | Iustin Pop | |
272 | 0e3baaf3 | Iustin Pop | if not os.path.exists(file_storage_dir): |
273 | 0e3baaf3 | Iustin Pop | try:
|
274 | 0e3baaf3 | Iustin Pop | os.makedirs(file_storage_dir, 0750)
|
275 | 0e3baaf3 | Iustin Pop | except OSError, err: |
276 | 0e3baaf3 | Iustin Pop | raise errors.OpPrereqError("Cannot create file storage directory" |
277 | 0e3baaf3 | Iustin Pop | " '%s': %s" % (file_storage_dir, err),
|
278 | 0e3baaf3 | Iustin Pop | errors.ECODE_ENVIRON) |
279 | 0e3baaf3 | Iustin Pop | |
280 | 0e3baaf3 | Iustin Pop | if not os.path.isdir(file_storage_dir): |
281 | 0e3baaf3 | Iustin Pop | raise errors.OpPrereqError("The file storage directory '%s' is not" |
282 | 0e3baaf3 | Iustin Pop | " a directory." % file_storage_dir,
|
283 | 0e3baaf3 | Iustin Pop | errors.ECODE_ENVIRON) |
284 | 0e3baaf3 | Iustin Pop | return file_storage_dir
|
285 | 0e3baaf3 | Iustin Pop | |
286 | 0e3baaf3 | Iustin Pop | |
287 | 18bb6d28 | Agata Murawska | def InitCluster(cluster_name, mac_prefix, # pylint: disable=R0913, R0914 |
288 | 5a8648eb | Andrea Spadaccini | master_netmask, master_netdev, file_storage_dir, |
289 | 5a8648eb | Andrea Spadaccini | shared_file_storage_dir, candidate_pool_size, secondary_ip=None,
|
290 | 5a8648eb | Andrea Spadaccini | vg_name=None, beparams=None, nicparams=None, ndparams=None, |
291 | bc5d0215 | Andrea Spadaccini | hvparams=None, diskparams=None, enabled_hypervisors=None, |
292 | bc5d0215 | Andrea Spadaccini | modify_etc_hosts=True, modify_ssh_setup=True, |
293 | bc5d0215 | Andrea Spadaccini | maintain_node_health=False, drbd_helper=None, uid_pool=None, |
294 | 18bb6d28 | Agata Murawska | default_iallocator=None, primary_ip_version=None, ipolicy=None, |
295 | c4929a8b | Renรฉ Nussbaumer | prealloc_wipe_disks=False, use_external_mip_script=False, |
296 | c4929a8b | Renรฉ Nussbaumer | hv_state=None, disk_state=None): |
297 | a0c9f010 | Michael Hanselmann | """Initialise the cluster.
|
298 | a0c9f010 | Michael Hanselmann |
|
299 | ce735215 | Guido Trotter | @type candidate_pool_size: int
|
300 | ce735215 | Guido Trotter | @param candidate_pool_size: master candidate pool size
|
301 | ce735215 | Guido Trotter |
|
302 | a0c9f010 | Michael Hanselmann | """
|
303 | ce735215 | Guido Trotter | # TODO: complete the docstring
|
304 | a0c9f010 | Michael Hanselmann | if config.ConfigWriter.IsCluster():
|
305 | debac808 | Iustin Pop | raise errors.OpPrereqError("Cluster is already initialised", |
306 | debac808 | Iustin Pop | errors.ECODE_STATE) |
307 | a0c9f010 | Michael Hanselmann | |
308 | b119bccb | Guido Trotter | if not enabled_hypervisors: |
309 | b119bccb | Guido Trotter | raise errors.OpPrereqError("Enabled hypervisors list must contain at" |
310 | debac808 | Iustin Pop | " least one member", errors.ECODE_INVAL)
|
311 | b119bccb | Guido Trotter | invalid_hvs = set(enabled_hypervisors) - constants.HYPER_TYPES
|
312 | b119bccb | Guido Trotter | if invalid_hvs:
|
313 | b119bccb | Guido Trotter | raise errors.OpPrereqError("Enabled hypervisors contains invalid" |
314 | debac808 | Iustin Pop | " entries: %s" % invalid_hvs,
|
315 | debac808 | Iustin Pop | errors.ECODE_INVAL) |
316 | b119bccb | Guido Trotter | |
317 | 5a8648eb | Andrea Spadaccini | try:
|
318 | 5a8648eb | Andrea Spadaccini | ipcls = netutils.IPAddress.GetClassFromIpVersion(primary_ip_version) |
319 | 5a8648eb | Andrea Spadaccini | except errors.ProgrammerError:
|
320 | 2f20d07b | Manuel Franceschini | raise errors.OpPrereqError("Invalid primary ip version: %d." % |
321 | 2f20d07b | Manuel Franceschini | primary_ip_version) |
322 | 2f20d07b | Manuel Franceschini | |
323 | 2f20d07b | Manuel Franceschini | hostname = netutils.GetHostname(family=ipcls.family) |
324 | 2f20d07b | Manuel Franceschini | if not ipcls.IsValid(hostname.ip): |
325 | 2f20d07b | Manuel Franceschini | raise errors.OpPrereqError("This host's IP (%s) is not a valid IPv%d" |
326 | 2f20d07b | Manuel Franceschini | " address." % (hostname.ip, primary_ip_version))
|
327 | 2f20d07b | Manuel Franceschini | |
328 | 2f20d07b | Manuel Franceschini | if ipcls.IsLoopback(hostname.ip):
|
329 | 8b312c1d | Manuel Franceschini | raise errors.OpPrereqError("This host's IP (%s) resolves to a loopback" |
330 | 8b312c1d | Manuel Franceschini | " address. Please fix DNS or %s." %
|
331 | debac808 | Iustin Pop | (hostname.ip, constants.ETC_HOSTS), |
332 | debac808 | Iustin Pop | errors.ECODE_ENVIRON) |
333 | a0c9f010 | Michael Hanselmann | |
334 | 2f20d07b | Manuel Franceschini | if not ipcls.Own(hostname.ip): |
335 | a0c9f010 | Michael Hanselmann | raise errors.OpPrereqError("Inconsistency: this host's name resolves" |
336 | a0c9f010 | Michael Hanselmann | " to %s,\nbut this ip address does not"
|
337 | 7c4c22f5 | Manuel Franceschini | " belong to this host" %
|
338 | debac808 | Iustin Pop | hostname.ip, errors.ECODE_ENVIRON) |
339 | a0c9f010 | Michael Hanselmann | |
340 | 2f20d07b | Manuel Franceschini | clustername = netutils.GetHostname(name=cluster_name, family=ipcls.family) |
341 | a0c9f010 | Michael Hanselmann | |
342 | 2f20d07b | Manuel Franceschini | if netutils.TcpPing(clustername.ip, constants.DEFAULT_NODED_PORT, timeout=5): |
343 | 7c4c22f5 | Manuel Franceschini | raise errors.OpPrereqError("Cluster IP already active", |
344 | debac808 | Iustin Pop | errors.ECODE_NOTUNIQUE) |
345 | a0c9f010 | Michael Hanselmann | |
346 | 2f20d07b | Manuel Franceschini | if not secondary_ip: |
347 | 2f20d07b | Manuel Franceschini | if primary_ip_version == constants.IP6_VERSION:
|
348 | 2f20d07b | Manuel Franceschini | raise errors.OpPrereqError("When using a IPv6 primary address, a valid" |
349 | 7c4c22f5 | Manuel Franceschini | " IPv4 address must be given as secondary",
|
350 | 7c4c22f5 | Manuel Franceschini | errors.ECODE_INVAL) |
351 | b9eeeb02 | Michael Hanselmann | secondary_ip = hostname.ip |
352 | a0c9f010 | Michael Hanselmann | |
353 | 2f20d07b | Manuel Franceschini | if not netutils.IP4Address.IsValid(secondary_ip): |
354 | 2f20d07b | Manuel Franceschini | raise errors.OpPrereqError("Secondary IP address (%s) has to be a valid" |
355 | 2f20d07b | Manuel Franceschini | " IPv4 address." % secondary_ip,
|
356 | 2f20d07b | Manuel Franceschini | errors.ECODE_INVAL) |
357 | 2f20d07b | Manuel Franceschini | |
358 | 2f20d07b | Manuel Franceschini | if not netutils.IP4Address.Own(secondary_ip): |
359 | 2f20d07b | Manuel Franceschini | raise errors.OpPrereqError("You gave %s as secondary IP," |
360 | 2f20d07b | Manuel Franceschini | " but it does not belong to this host." %
|
361 | 2f20d07b | Manuel Franceschini | secondary_ip, errors.ECODE_ENVIRON) |
362 | 2f20d07b | Manuel Franceschini | |
363 | 5a8648eb | Andrea Spadaccini | if master_netmask is not None: |
364 | 5a8648eb | Andrea Spadaccini | if not ipcls.ValidateNetmask(master_netmask): |
365 | 5a8648eb | Andrea Spadaccini | raise errors.OpPrereqError("CIDR netmask (%s) not valid for IPv%s " % |
366 | 5a8648eb | Andrea Spadaccini | (master_netmask, primary_ip_version)) |
367 | 5a8648eb | Andrea Spadaccini | else:
|
368 | 5a8648eb | Andrea Spadaccini | master_netmask = ipcls.iplen |
369 | 5a8648eb | Andrea Spadaccini | |
370 | a0c9f010 | Michael Hanselmann | if vg_name is not None: |
371 | a0c9f010 | Michael Hanselmann | # Check if volume group is valid
|
372 | a0c9f010 | Michael Hanselmann | vgstatus = utils.CheckVolumeGroupSize(utils.ListVolumeGroups(), vg_name, |
373 | a0c9f010 | Michael Hanselmann | constants.MIN_VG_SIZE) |
374 | a0c9f010 | Michael Hanselmann | if vgstatus:
|
375 | a0c9f010 | Michael Hanselmann | raise errors.OpPrereqError("Error: %s\nspecify --no-lvm-storage if" |
376 | debac808 | Iustin Pop | " you are not using lvm" % vgstatus,
|
377 | debac808 | Iustin Pop | errors.ECODE_INVAL) |
378 | a0c9f010 | Michael Hanselmann | |
379 | a721e23a | Luca Bigliardi | if drbd_helper is not None: |
380 | a721e23a | Luca Bigliardi | try:
|
381 | a721e23a | Luca Bigliardi | curr_helper = bdev.BaseDRBD.GetUsermodeHelper() |
382 | a721e23a | Luca Bigliardi | except errors.BlockDeviceError, err:
|
383 | a721e23a | Luca Bigliardi | raise errors.OpPrereqError("Error while checking drbd helper" |
384 | a721e23a | Luca Bigliardi | " (specify --no-drbd-storage if you are not"
|
385 | a721e23a | Luca Bigliardi | " using drbd): %s" % str(err), |
386 | a721e23a | Luca Bigliardi | errors.ECODE_ENVIRON) |
387 | a721e23a | Luca Bigliardi | if drbd_helper != curr_helper:
|
388 | a721e23a | Luca Bigliardi | raise errors.OpPrereqError("Error: requiring %s as drbd helper but %s" |
389 | a721e23a | Luca Bigliardi | " is the current helper" % (drbd_helper,
|
390 | a721e23a | Luca Bigliardi | curr_helper), |
391 | a721e23a | Luca Bigliardi | errors.ECODE_INVAL) |
392 | a721e23a | Luca Bigliardi | |
393 | 0376655e | Guido Trotter | if constants.ENABLE_FILE_STORAGE:
|
394 | 0376655e | Guido Trotter | file_storage_dir = _InitFileStorage(file_storage_dir) |
395 | 0376655e | Guido Trotter | else:
|
396 | 0376655e | Guido Trotter | file_storage_dir = ""
|
397 | 0376655e | Guido Trotter | |
398 | 0376655e | Guido Trotter | if constants.ENABLE_SHARED_FILE_STORAGE:
|
399 | 0376655e | Guido Trotter | shared_file_storage_dir = _InitFileStorage(shared_file_storage_dir) |
400 | 0376655e | Guido Trotter | else:
|
401 | 0376655e | Guido Trotter | shared_file_storage_dir = ""
|
402 | a0c9f010 | Michael Hanselmann | |
403 | a0c9f010 | Michael Hanselmann | if not re.match("^[0-9a-z]{2}:[0-9a-z]{2}:[0-9a-z]{2}$", mac_prefix): |
404 | debac808 | Iustin Pop | raise errors.OpPrereqError("Invalid mac prefix given '%s'" % mac_prefix, |
405 | debac808 | Iustin Pop | errors.ECODE_INVAL) |
406 | a0c9f010 | Michael Hanselmann | |
407 | a0c9f010 | Michael Hanselmann | result = utils.RunCmd(["ip", "link", "show", "dev", master_netdev]) |
408 | a0c9f010 | Michael Hanselmann | if result.failed:
|
409 | a0c9f010 | Michael Hanselmann | raise errors.OpPrereqError("Invalid master netdev given (%s): '%s'" % |
410 | a0c9f010 | Michael Hanselmann | (master_netdev, |
411 | debac808 | Iustin Pop | result.output.strip()), errors.ECODE_INVAL) |
412 | a0c9f010 | Michael Hanselmann | |
413 | 9dae41ad | Guido Trotter | dirs = [(constants.RUN_GANETI_DIR, constants.RUN_DIRS_MODE)] |
414 | 9dae41ad | Guido Trotter | utils.EnsureDirs(dirs) |
415 | 9dae41ad | Guido Trotter | |
416 | b2e233a5 | Guido Trotter | objects.UpgradeBeParams(beparams) |
417 | a5728081 | Guido Trotter | utils.ForceDictType(beparams, constants.BES_PARAMETER_TYPES) |
418 | b6a30b0d | Guido Trotter | utils.ForceDictType(nicparams, constants.NICS_PARAMETER_TYPES) |
419 | 18bb6d28 | Agata Murawska | |
420 | b6a30b0d | Guido Trotter | objects.NIC.CheckParameterSyntax(nicparams) |
421 | 57dc299a | Iustin Pop | |
422 | 2cc673a3 | Iustin Pop | full_ipolicy = objects.FillIPolicy(constants.IPOLICY_DEFAULTS, ipolicy) |
423 | b6a30b0d | Guido Trotter | |
424 | 6204ee71 | Renรฉ Nussbaumer | if ndparams is not None: |
425 | 6204ee71 | Renรฉ Nussbaumer | utils.ForceDictType(ndparams, constants.NDS_PARAMETER_TYPES) |
426 | 6204ee71 | Renรฉ Nussbaumer | else:
|
427 | 6204ee71 | Renรฉ Nussbaumer | ndparams = dict(constants.NDC_DEFAULTS)
|
428 | 6204ee71 | Renรฉ Nussbaumer | |
429 | c4929a8b | Renรฉ Nussbaumer | # This is ugly, as we modify the dict itself
|
430 | 57dc299a | Iustin Pop | # FIXME: Make utils.ForceDictType pure functional or write a wrapper
|
431 | 57dc299a | Iustin Pop | # around it
|
432 | c4929a8b | Renรฉ Nussbaumer | if hv_state:
|
433 | c4929a8b | Renรฉ Nussbaumer | for hvname, hvs_data in hv_state.items(): |
434 | c4929a8b | Renรฉ Nussbaumer | utils.ForceDictType(hvs_data, constants.HVSTS_PARAMETER_TYPES) |
435 | c4929a8b | Renรฉ Nussbaumer | hv_state[hvname] = objects.Cluster.SimpleFillHvState(hvs_data) |
436 | c4929a8b | Renรฉ Nussbaumer | else:
|
437 | c4929a8b | Renรฉ Nussbaumer | hv_state = dict((hvname, constants.HVST_DEFAULTS)
|
438 | c4929a8b | Renรฉ Nussbaumer | for hvname in enabled_hypervisors) |
439 | c4929a8b | Renรฉ Nussbaumer | |
440 | c4929a8b | Renรฉ Nussbaumer | # FIXME: disk_state has no default values yet
|
441 | c4929a8b | Renรฉ Nussbaumer | if disk_state:
|
442 | c4929a8b | Renรฉ Nussbaumer | for storage, ds_data in disk_state.items(): |
443 | c4929a8b | Renรฉ Nussbaumer | if storage not in constants.DS_VALID_TYPES: |
444 | c4929a8b | Renรฉ Nussbaumer | raise errors.OpPrereqError("Invalid storage type in disk state: %s" % |
445 | c4929a8b | Renรฉ Nussbaumer | storage, errors.ECODE_INVAL) |
446 | c4929a8b | Renรฉ Nussbaumer | for ds_name, state in ds_data.items(): |
447 | c4929a8b | Renรฉ Nussbaumer | utils.ForceDictType(state, constants.DSS_PARAMETER_TYPES) |
448 | c4929a8b | Renรฉ Nussbaumer | ds_data[ds_name] = objects.Cluster.SimpleFillDiskState(state) |
449 | c4929a8b | Renรฉ Nussbaumer | |
450 | a5728081 | Guido Trotter | # hvparams is a mapping of hypervisor->hvparams dict
|
451 | a5728081 | Guido Trotter | for hv_name, hv_params in hvparams.iteritems(): |
452 | a5728081 | Guido Trotter | utils.ForceDictType(hv_params, constants.HVS_PARAMETER_TYPES) |
453 | a5728081 | Guido Trotter | hv_class = hypervisor.GetHypervisor(hv_name) |
454 | a5728081 | Guido Trotter | hv_class.CheckParameterSyntax(hv_params) |
455 | d4b72030 | Guido Trotter | |
456 | bc5d0215 | Andrea Spadaccini | # diskparams is a mapping of disk-template->diskparams dict
|
457 | bc5d0215 | Andrea Spadaccini | for template, dt_params in diskparams.items(): |
458 | bc5d0215 | Andrea Spadaccini | param_keys = set(dt_params.keys())
|
459 | bc5d0215 | Andrea Spadaccini | default_param_keys = set(constants.DISK_DT_DEFAULTS[template].keys())
|
460 | bc5d0215 | Andrea Spadaccini | if not (param_keys <= default_param_keys): |
461 | bc5d0215 | Andrea Spadaccini | unknown_params = param_keys - default_param_keys |
462 | bc5d0215 | Andrea Spadaccini | raise errors.OpPrereqError("Invalid parameters for disk template %s:" |
463 | bc5d0215 | Andrea Spadaccini | " %s" % (template,
|
464 | bc5d0215 | Andrea Spadaccini | utils.CommaJoin(unknown_params))) |
465 | bc5d0215 | Andrea Spadaccini | utils.ForceDictType(dt_params, constants.DISK_DT_TYPES) |
466 | e4a4391d | Renรฉ Nussbaumer | try:
|
467 | e4a4391d | Renรฉ Nussbaumer | utils.VerifyDictOptions(diskparams, constants.DISK_DT_DEFAULTS) |
468 | e4a4391d | Renรฉ Nussbaumer | except errors.OpPrereqError, err:
|
469 | e4a4391d | Renรฉ Nussbaumer | raise errors.OpPrereqError("While verify diskparam options: %s" % err, |
470 | e4a4391d | Renรฉ Nussbaumer | errors.ECODE_INVAL) |
471 | bc5d0215 | Andrea Spadaccini | |
472 | a0c9f010 | Michael Hanselmann | # set up ssh config and /etc/hosts
|
473 | 13998ef2 | Michael Hanselmann | sshline = utils.ReadFile(constants.SSH_HOST_RSA_PUB) |
474 | a0c9f010 | Michael Hanselmann | sshkey = sshline.split(" ")[1] |
475 | a0c9f010 | Michael Hanselmann | |
476 | b86a6bcd | Guido Trotter | if modify_etc_hosts:
|
477 | ea8ac9c9 | Renรฉ Nussbaumer | utils.AddHostToEtcHosts(hostname.name, hostname.ip) |
478 | b86a6bcd | Guido Trotter | |
479 | b989b9d9 | Ken Wehr | if modify_ssh_setup:
|
480 | b989b9d9 | Ken Wehr | _InitSSHSetup() |
481 | a0c9f010 | Michael Hanselmann | |
482 | bf4af505 | Apollon Oikonomopoulos | if default_iallocator is not None: |
483 | bf4af505 | Apollon Oikonomopoulos | alloc_script = utils.FindFile(default_iallocator, |
484 | bf4af505 | Apollon Oikonomopoulos | constants.IALLOCATOR_SEARCH_PATH, |
485 | bf4af505 | Apollon Oikonomopoulos | os.path.isfile) |
486 | bf4af505 | Apollon Oikonomopoulos | if alloc_script is None: |
487 | bf4af505 | Apollon Oikonomopoulos | raise errors.OpPrereqError("Invalid default iallocator script '%s'" |
488 | bf4af505 | Apollon Oikonomopoulos | " specified" % default_iallocator,
|
489 | bf4af505 | Apollon Oikonomopoulos | errors.ECODE_INVAL) |
490 | d1e9c98d | Iustin Pop | elif constants.HTOOLS:
|
491 | d1e9c98d | Iustin Pop | # htools was enabled at build-time, we default to it
|
492 | d1e9c98d | Iustin Pop | if utils.FindFile(constants.IALLOC_HAIL,
|
493 | d1e9c98d | Iustin Pop | constants.IALLOCATOR_SEARCH_PATH, |
494 | d1e9c98d | Iustin Pop | os.path.isfile): |
495 | d1e9c98d | Iustin Pop | default_iallocator = constants.IALLOC_HAIL |
496 | bf4af505 | Apollon Oikonomopoulos | |
497 | 430b923c | Iustin Pop | now = time.time() |
498 | 430b923c | Iustin Pop | |
499 | a0c9f010 | Michael Hanselmann | # init of cluster config file
|
500 | b9eeeb02 | Michael Hanselmann | cluster_config = objects.Cluster( |
501 | b9eeeb02 | Michael Hanselmann | serial_no=1,
|
502 | b9eeeb02 | Michael Hanselmann | rsahostkeypub=sshkey, |
503 | b9eeeb02 | Michael Hanselmann | highest_used_port=(constants.FIRST_DRBD_PORT - 1),
|
504 | b9eeeb02 | Michael Hanselmann | mac_prefix=mac_prefix, |
505 | b9eeeb02 | Michael Hanselmann | volume_group_name=vg_name, |
506 | b9eeeb02 | Michael Hanselmann | tcpudp_port_pool=set(),
|
507 | f6bd6e98 | Michael Hanselmann | master_node=hostname.name, |
508 | f6bd6e98 | Michael Hanselmann | master_ip=clustername.ip, |
509 | 5a8648eb | Andrea Spadaccini | master_netmask=master_netmask, |
510 | f6bd6e98 | Michael Hanselmann | master_netdev=master_netdev, |
511 | f6bd6e98 | Michael Hanselmann | cluster_name=clustername.name, |
512 | f6bd6e98 | Michael Hanselmann | file_storage_dir=file_storage_dir, |
513 | 4b97f902 | Apollon Oikonomopoulos | shared_file_storage_dir=shared_file_storage_dir, |
514 | ea3a925f | Alexander Schreiber | enabled_hypervisors=enabled_hypervisors, |
515 | 4ef7f423 | Guido Trotter | beparams={constants.PP_DEFAULT: beparams}, |
516 | b6a30b0d | Guido Trotter | nicparams={constants.PP_DEFAULT: nicparams}, |
517 | 6204ee71 | Renรฉ Nussbaumer | ndparams=ndparams, |
518 | ea3a925f | Alexander Schreiber | hvparams=hvparams, |
519 | bc5d0215 | Andrea Spadaccini | diskparams=diskparams, |
520 | ce735215 | Guido Trotter | candidate_pool_size=candidate_pool_size, |
521 | 022c3a0b | Guido Trotter | modify_etc_hosts=modify_etc_hosts, |
522 | b989b9d9 | Ken Wehr | modify_ssh_setup=modify_ssh_setup, |
523 | 39b0f0c2 | Balazs Lecz | uid_pool=uid_pool, |
524 | 430b923c | Iustin Pop | ctime=now, |
525 | 430b923c | Iustin Pop | mtime=now, |
526 | 3953242f | Iustin Pop | maintain_node_health=maintain_node_health, |
527 | a721e23a | Luca Bigliardi | drbd_usermode_helper=drbd_helper, |
528 | bf4af505 | Apollon Oikonomopoulos | default_iallocator=default_iallocator, |
529 | 2f20d07b | Manuel Franceschini | primary_ip_family=ipcls.family, |
530 | 3d914585 | Renรฉ Nussbaumer | prealloc_wipe_disks=prealloc_wipe_disks, |
531 | bf689b7a | Andrea Spadaccini | use_external_mip_script=use_external_mip_script, |
532 | 57dc299a | Iustin Pop | ipolicy=full_ipolicy, |
533 | c4929a8b | Renรฉ Nussbaumer | hv_state_static=hv_state, |
534 | c4929a8b | Renรฉ Nussbaumer | disk_state_static=disk_state, |
535 | b9eeeb02 | Michael Hanselmann | ) |
536 | b9eeeb02 | Michael Hanselmann | master_node_config = objects.Node(name=hostname.name, |
537 | b9eeeb02 | Michael Hanselmann | primary_ip=hostname.ip, |
538 | b9222f32 | Guido Trotter | secondary_ip=secondary_ip, |
539 | c044f32c | Guido Trotter | serial_no=1,
|
540 | c044f32c | Guido Trotter | master_candidate=True,
|
541 | af64c0ea | Iustin Pop | offline=False, drained=False, |
542 | 435e4bd6 | Michael Hanselmann | ctime=now, mtime=now, |
543 | c044f32c | Guido Trotter | ) |
544 | 9e1333b9 | Guido Trotter | InitConfig(constants.CONFIG_VERSION, cluster_config, master_node_config) |
545 | d367b66c | Manuel Franceschini | cfg = config.ConfigWriter(offline=True)
|
546 | 9e1333b9 | Guido Trotter | ssh.WriteKnownHostsFile(cfg, constants.SSH_KNOWN_HOSTS_FILE) |
547 | a4eae71f | Michael Hanselmann | cfg.Update(cfg.GetClusterInfo(), logging.error) |
548 | d367b66c | Manuel Franceschini | backend.WriteSsconfFiles(cfg.GetSsconfValues()) |
549 | d367b66c | Manuel Franceschini | |
550 | d367b66c | Manuel Franceschini | # set up the inter-node password and certificate
|
551 | d367b66c | Manuel Franceschini | _InitGanetiServerSetup(hostname.name) |
552 | 827f753e | Guido Trotter | |
553 | 952d7515 | Michael Hanselmann | logging.debug("Starting daemons")
|
554 | 952d7515 | Michael Hanselmann | result = utils.RunCmd([constants.DAEMON_UTIL, "start-all"])
|
555 | 952d7515 | Michael Hanselmann | if result.failed:
|
556 | 952d7515 | Michael Hanselmann | raise errors.OpExecError("Could not start daemons, command %s" |
557 | 952d7515 | Michael Hanselmann | " had exitcode %s and error %s" %
|
558 | 952d7515 | Michael Hanselmann | (result.cmd, result.exit_code, result.output)) |
559 | b3f1cf6f | Iustin Pop | |
560 | 3b6b6129 | Michael Hanselmann | _WaitForMasterDaemon() |
561 | b3f1cf6f | Iustin Pop | |
562 | b1b6ea87 | Iustin Pop | |
563 | 02f99608 | Oleksiy Mishchenko | def InitConfig(version, cluster_config, master_node_config, |
564 | 02f99608 | Oleksiy Mishchenko | cfg_file=constants.CLUSTER_CONF_FILE): |
565 | 7b3a8fb5 | Iustin Pop | """Create the initial cluster configuration.
|
566 | 7b3a8fb5 | Iustin Pop |
|
567 | 7b3a8fb5 | Iustin Pop | It will contain the current node, which will also be the master
|
568 | 7b3a8fb5 | Iustin Pop | node, and no instances.
|
569 | 7b3a8fb5 | Iustin Pop |
|
570 | 7b3a8fb5 | Iustin Pop | @type version: int
|
571 | c41eea6e | Iustin Pop | @param version: configuration version
|
572 | c41eea6e | Iustin Pop | @type cluster_config: L{objects.Cluster}
|
573 | c41eea6e | Iustin Pop | @param cluster_config: cluster configuration
|
574 | c41eea6e | Iustin Pop | @type master_node_config: L{objects.Node}
|
575 | c41eea6e | Iustin Pop | @param master_node_config: master node configuration
|
576 | c41eea6e | Iustin Pop | @type cfg_file: string
|
577 | c41eea6e | Iustin Pop | @param cfg_file: configuration file path
|
578 | c41eea6e | Iustin Pop |
|
579 | 7b3a8fb5 | Iustin Pop | """
|
580 | 88b92fe3 | Guido Trotter | uuid_generator = config.TemporaryReservationManager() |
581 | 88b92fe3 | Guido Trotter | cluster_config.uuid = uuid_generator.Generate([], utils.NewUUID, |
582 | 88b92fe3 | Guido Trotter | _INITCONF_ECID) |
583 | 88b92fe3 | Guido Trotter | master_node_config.uuid = uuid_generator.Generate([], utils.NewUUID, |
584 | 88b92fe3 | Guido Trotter | _INITCONF_ECID) |
585 | 7b3a8fb5 | Iustin Pop | nodes = { |
586 | 7b3a8fb5 | Iustin Pop | master_node_config.name: master_node_config, |
587 | 7b3a8fb5 | Iustin Pop | } |
588 | 88b92fe3 | Guido Trotter | default_nodegroup = objects.NodeGroup( |
589 | 88b92fe3 | Guido Trotter | uuid=uuid_generator.Generate([], utils.NewUUID, _INITCONF_ECID), |
590 | 75cf411a | Adeodato Simo | name=constants.INITIAL_NODE_GROUP_NAME, |
591 | 88b92fe3 | Guido Trotter | members=[master_node_config.name], |
592 | 99ccf8b9 | Renรฉ Nussbaumer | diskparams={}, |
593 | 88b92fe3 | Guido Trotter | ) |
594 | 88b92fe3 | Guido Trotter | nodegroups = { |
595 | 88b92fe3 | Guido Trotter | default_nodegroup.uuid: default_nodegroup, |
596 | 88b92fe3 | Guido Trotter | } |
597 | d693c864 | Iustin Pop | now = time.time() |
598 | 7b3a8fb5 | Iustin Pop | config_data = objects.ConfigData(version=version, |
599 | 7b3a8fb5 | Iustin Pop | cluster=cluster_config, |
600 | 88b92fe3 | Guido Trotter | nodegroups=nodegroups, |
601 | 7b3a8fb5 | Iustin Pop | nodes=nodes, |
602 | 7b3a8fb5 | Iustin Pop | instances={}, |
603 | f4f36cf3 | Dimitris Aragiorgis | networks={}, |
604 | d693c864 | Iustin Pop | serial_no=1,
|
605 | d693c864 | Iustin Pop | ctime=now, mtime=now) |
606 | a33848a5 | Guido Trotter | utils.WriteFile(cfg_file, |
607 | a33848a5 | Guido Trotter | data=serializer.Dump(config_data.ToDict()), |
608 | a33848a5 | Guido Trotter | mode=0600)
|
609 | 02f99608 | Oleksiy Mishchenko | |
610 | 02f99608 | Oleksiy Mishchenko | |
611 | 140aa4a8 | Iustin Pop | def FinalizeClusterDestroy(master): |
612 | 140aa4a8 | Iustin Pop | """Execute the last steps of cluster destroy
|
613 | 140aa4a8 | Iustin Pop |
|
614 | 140aa4a8 | Iustin Pop | This function shuts down all the daemons, completing the destroy
|
615 | 140aa4a8 | Iustin Pop | begun in cmdlib.LUDestroyOpcode.
|
616 | 140aa4a8 | Iustin Pop |
|
617 | 140aa4a8 | Iustin Pop | """
|
618 | b989b9d9 | Ken Wehr | cfg = config.ConfigWriter() |
619 | b989b9d9 | Ken Wehr | modify_ssh_setup = cfg.GetClusterInfo().modify_ssh_setup |
620 | 7c74bbe0 | Andrea Spadaccini | runner = rpc.BootstrapRunner() |
621 | 7c74bbe0 | Andrea Spadaccini | |
622 | f9d20654 | Andrea Spadaccini | master_params = cfg.GetMasterNetworkParameters() |
623 | c79198a0 | Andrea Spadaccini | master_params.name = master |
624 | 57c7bc57 | Andrea Spadaccini | ems = cfg.GetUseExternalMipScript() |
625 | c79198a0 | Andrea Spadaccini | result = runner.call_node_deactivate_master_ip(master_params.name, |
626 | 57c7bc57 | Andrea Spadaccini | master_params, ems) |
627 | c79198a0 | Andrea Spadaccini | |
628 | 7c74bbe0 | Andrea Spadaccini | msg = result.fail_msg |
629 | 7c74bbe0 | Andrea Spadaccini | if msg:
|
630 | 7c74bbe0 | Andrea Spadaccini | logging.warning("Could not disable the master IP: %s", msg)
|
631 | 7c74bbe0 | Andrea Spadaccini | |
632 | 7c74bbe0 | Andrea Spadaccini | result = runner.call_node_stop_master(master) |
633 | 3cebe102 | Michael Hanselmann | msg = result.fail_msg |
634 | 6c00d19a | Iustin Pop | if msg:
|
635 | 099c52ad | Iustin Pop | logging.warning("Could not disable the master role: %s", msg)
|
636 | 7c74bbe0 | Andrea Spadaccini | |
637 | 7c74bbe0 | Andrea Spadaccini | result = runner.call_node_leave_cluster(master, modify_ssh_setup) |
638 | 3cebe102 | Michael Hanselmann | msg = result.fail_msg |
639 | 0623d351 | Iustin Pop | if msg:
|
640 | 0623d351 | Iustin Pop | logging.warning("Could not shutdown the node daemon and cleanup"
|
641 | 0623d351 | Iustin Pop | " the node: %s", msg)
|
642 | 140aa4a8 | Iustin Pop | |
643 | 140aa4a8 | Iustin Pop | |
644 | 87622829 | Iustin Pop | def SetupNodeDaemon(cluster_name, node, ssh_key_check): |
645 | 827f753e | Guido Trotter | """Add a node to the cluster.
|
646 | 827f753e | Guido Trotter |
|
647 | b1b6ea87 | Iustin Pop | This function must be called before the actual opcode, and will ssh
|
648 | b1b6ea87 | Iustin Pop | to the remote node, copy the needed files, and start ganeti-noded,
|
649 | b1b6ea87 | Iustin Pop | allowing the master to do the rest via normal rpc calls.
|
650 | 827f753e | Guido Trotter |
|
651 | 87622829 | Iustin Pop | @param cluster_name: the cluster name
|
652 | 87622829 | Iustin Pop | @param node: the name of the new node
|
653 | 87622829 | Iustin Pop | @param ssh_key_check: whether to do a strict key check
|
654 | 827f753e | Guido Trotter |
|
655 | 827f753e | Guido Trotter | """
|
656 | b43dcc5a | Manuel Franceschini | family = ssconf.SimpleStore().GetPrimaryIPFamily() |
657 | b43dcc5a | Manuel Franceschini | sshrunner = ssh.SshRunner(cluster_name, |
658 | 72729d6e | Michael Hanselmann | ipv6=(family == netutils.IP6Address.family)) |
659 | 827f753e | Guido Trotter | |
660 | b43dcc5a | Manuel Franceschini | bind_address = constants.IP4_ADDRESS_ANY |
661 | b43dcc5a | Manuel Franceschini | if family == netutils.IP6Address.family:
|
662 | b43dcc5a | Manuel Franceschini | bind_address = constants.IP6_ADDRESS_ANY |
663 | b43dcc5a | Manuel Franceschini | |
664 | 827f753e | Guido Trotter | # set up inter-node password and certificate and restarts the node daemon
|
665 | 827f753e | Guido Trotter | # and then connect with ssh to set password and start ganeti-noded
|
666 | 827f753e | Guido Trotter | # note that all the below variables are sanitized at this point,
|
667 | 827f753e | Guido Trotter | # either by being constants or by the checks above
|
668 | 9294514d | Renรฉ Nussbaumer | sshrunner.CopyFileToNode(node, constants.NODED_CERT_FILE) |
669 | 9294514d | Renรฉ Nussbaumer | sshrunner.CopyFileToNode(node, constants.RAPI_CERT_FILE) |
670 | bfe86c76 | Andrea Spadaccini | sshrunner.CopyFileToNode(node, constants.SPICE_CERT_FILE) |
671 | bfe86c76 | Andrea Spadaccini | sshrunner.CopyFileToNode(node, constants.SPICE_CACERT_FILE) |
672 | 9294514d | Renรฉ Nussbaumer | sshrunner.CopyFileToNode(node, constants.CONFD_HMAC_KEY) |
673 | 72729d6e | Michael Hanselmann | mycommand = ("%s stop-all; %s start %s -b %s" %
|
674 | 72729d6e | Michael Hanselmann | (constants.DAEMON_UTIL, constants.DAEMON_UTIL, constants.NODED, |
675 | 72729d6e | Michael Hanselmann | utils.ShellQuote(bind_address))) |
676 | 827f753e | Guido Trotter | |
677 | 3ccb3a64 | Michael Hanselmann | result = sshrunner.Run(node, "root", mycommand, batch=False, |
678 | c4b6c29c | Michael Hanselmann | ask_key=ssh_key_check, |
679 | 9294514d | Renรฉ Nussbaumer | use_cluster_key=True,
|
680 | c4b6c29c | Michael Hanselmann | strict_host_check=ssh_key_check) |
681 | 827f753e | Guido Trotter | if result.failed:
|
682 | 827f753e | Guido Trotter | raise errors.OpExecError("Remote command on node %s, error: %s," |
683 | 827f753e | Guido Trotter | " output: %s" %
|
684 | 827f753e | Guido Trotter | (node, result.fail_reason, result.output)) |
685 | 827f753e | Guido Trotter | |
686 | 5627f375 | Michael Hanselmann | _WaitForNodeDaemon(node) |
687 | 5627f375 | Michael Hanselmann | |
688 | b1b6ea87 | Iustin Pop | |
689 | 8e2524c3 | Guido Trotter | def MasterFailover(no_voting=False): |
690 | b1b6ea87 | Iustin Pop | """Failover the master node.
|
691 | b1b6ea87 | Iustin Pop |
|
692 | b1b6ea87 | Iustin Pop | This checks that we are not already the master, and will cause the
|
693 | b1b6ea87 | Iustin Pop | current master to cease being master, and the non-master to become
|
694 | b1b6ea87 | Iustin Pop | new master.
|
695 | b1b6ea87 | Iustin Pop |
|
696 | 8e2524c3 | Guido Trotter | @type no_voting: boolean
|
697 | 8e2524c3 | Guido Trotter | @param no_voting: force the operation without remote nodes agreement
|
698 | 8e2524c3 | Guido Trotter | (dangerous)
|
699 | 8e2524c3 | Guido Trotter |
|
700 | b1b6ea87 | Iustin Pop | """
|
701 | 8135a2db | Iustin Pop | sstore = ssconf.SimpleStore() |
702 | b1b6ea87 | Iustin Pop | |
703 | 8135a2db | Iustin Pop | old_master, new_master = ssconf.GetMasterAndMyself(sstore) |
704 | 8135a2db | Iustin Pop | node_list = sstore.GetNodeList() |
705 | 8135a2db | Iustin Pop | mc_list = sstore.GetMasterCandidates() |
706 | b1b6ea87 | Iustin Pop | |
707 | b1b6ea87 | Iustin Pop | if old_master == new_master:
|
708 | b1b6ea87 | Iustin Pop | raise errors.OpPrereqError("This commands must be run on the node" |
709 | b1b6ea87 | Iustin Pop | " where you want the new master to be."
|
710 | b1b6ea87 | Iustin Pop | " %s is already the master" %
|
711 | debac808 | Iustin Pop | old_master, errors.ECODE_INVAL) |
712 | d5927e48 | Iustin Pop | |
713 | 8135a2db | Iustin Pop | if new_master not in mc_list: |
714 | 8135a2db | Iustin Pop | mc_no_master = [name for name in mc_list if name != old_master] |
715 | 8135a2db | Iustin Pop | raise errors.OpPrereqError("This node is not among the nodes marked" |
716 | 8135a2db | Iustin Pop | " as master candidates. Only these nodes"
|
717 | 8135a2db | Iustin Pop | " can become masters. Current list of"
|
718 | 8135a2db | Iustin Pop | " master candidates is:\n"
|
719 | 3ccb3a64 | Michael Hanselmann | "%s" % ("\n".join(mc_no_master)), |
720 | debac808 | Iustin Pop | errors.ECODE_STATE) |
721 | 8135a2db | Iustin Pop | |
722 | 8e2524c3 | Guido Trotter | if not no_voting: |
723 | 8e2524c3 | Guido Trotter | vote_list = GatherMasterVotes(node_list) |
724 | 8e2524c3 | Guido Trotter | |
725 | 8e2524c3 | Guido Trotter | if vote_list:
|
726 | 8e2524c3 | Guido Trotter | voted_master = vote_list[0][0] |
727 | 8e2524c3 | Guido Trotter | if voted_master is None: |
728 | 8e2524c3 | Guido Trotter | raise errors.OpPrereqError("Cluster is inconsistent, most nodes did" |
729 | debac808 | Iustin Pop | " not respond.", errors.ECODE_ENVIRON)
|
730 | 8e2524c3 | Guido Trotter | elif voted_master != old_master:
|
731 | 8e2524c3 | Guido Trotter | raise errors.OpPrereqError("I have a wrong configuration, I believe" |
732 | 8e2524c3 | Guido Trotter | " the master is %s but the other nodes"
|
733 | 8e2524c3 | Guido Trotter | " voted %s. Please resync the configuration"
|
734 | 8e2524c3 | Guido Trotter | " of this node." %
|
735 | debac808 | Iustin Pop | (old_master, voted_master), |
736 | debac808 | Iustin Pop | errors.ECODE_STATE) |
737 | b1b6ea87 | Iustin Pop | # end checks
|
738 | b1b6ea87 | Iustin Pop | |
739 | b1b6ea87 | Iustin Pop | rcode = 0
|
740 | b1b6ea87 | Iustin Pop | |
741 | d5927e48 | Iustin Pop | logging.info("Setting master to %s, old master: %s", new_master, old_master)
|
742 | b1b6ea87 | Iustin Pop | |
743 | 21004460 | Iustin Pop | try:
|
744 | 21004460 | Iustin Pop | # instantiate a real config writer, as we now know we have the
|
745 | 21004460 | Iustin Pop | # configuration data
|
746 | eb180fe2 | Iustin Pop | cfg = config.ConfigWriter(accept_foreign=True)
|
747 | 21004460 | Iustin Pop | |
748 | 21004460 | Iustin Pop | cluster_info = cfg.GetClusterInfo() |
749 | 21004460 | Iustin Pop | cluster_info.master_node = new_master |
750 | 21004460 | Iustin Pop | # this will also regenerate the ssconf files, since we updated the
|
751 | 21004460 | Iustin Pop | # cluster info
|
752 | 21004460 | Iustin Pop | cfg.Update(cluster_info, logging.error) |
753 | 21004460 | Iustin Pop | except errors.ConfigurationError, err:
|
754 | 21004460 | Iustin Pop | logging.error("Error while trying to set the new master: %s",
|
755 | 21004460 | Iustin Pop | str(err))
|
756 | 21004460 | Iustin Pop | return 1 |
757 | 21004460 | Iustin Pop | |
758 | 21004460 | Iustin Pop | # if cfg.Update worked, then it means the old master daemon won't be
|
759 | 21004460 | Iustin Pop | # able now to write its own config file (we rely on locking in both
|
760 | 21004460 | Iustin Pop | # backend.UploadFile() and ConfigWriter._Write(); hence the next
|
761 | 21004460 | Iustin Pop | # step is to kill the old master
|
762 | 21004460 | Iustin Pop | |
763 | 21004460 | Iustin Pop | logging.info("Stopping the master daemon on node %s", old_master)
|
764 | 21004460 | Iustin Pop | |
765 | 7c74bbe0 | Andrea Spadaccini | runner = rpc.BootstrapRunner() |
766 | f9d20654 | Andrea Spadaccini | master_params = cfg.GetMasterNetworkParameters() |
767 | c79198a0 | Andrea Spadaccini | master_params.name = old_master |
768 | 57c7bc57 | Andrea Spadaccini | ems = cfg.GetUseExternalMipScript() |
769 | c79198a0 | Andrea Spadaccini | result = runner.call_node_deactivate_master_ip(master_params.name, |
770 | 57c7bc57 | Andrea Spadaccini | master_params, ems) |
771 | c79198a0 | Andrea Spadaccini | |
772 | 7c74bbe0 | Andrea Spadaccini | msg = result.fail_msg |
773 | 7c74bbe0 | Andrea Spadaccini | if msg:
|
774 | 7c74bbe0 | Andrea Spadaccini | logging.warning("Could not disable the master IP: %s", msg)
|
775 | 7c74bbe0 | Andrea Spadaccini | |
776 | 7c74bbe0 | Andrea Spadaccini | result = runner.call_node_stop_master(old_master) |
777 | 3cebe102 | Michael Hanselmann | msg = result.fail_msg |
778 | 6c00d19a | Iustin Pop | if msg:
|
779 | d5927e48 | Iustin Pop | logging.error("Could not disable the master role on the old master"
|
780 | 6c00d19a | Iustin Pop | " %s, please disable manually: %s", old_master, msg)
|
781 | b1b6ea87 | Iustin Pop | |
782 | 21004460 | Iustin Pop | logging.info("Checking master IP non-reachability...")
|
783 | 21004460 | Iustin Pop | |
784 | 425f0f54 | Iustin Pop | master_ip = sstore.GetMasterIP() |
785 | 425f0f54 | Iustin Pop | total_timeout = 30
|
786 | e687ec01 | Michael Hanselmann | |
787 | d23ef431 | Michael Hanselmann | # Here we have a phase where no master should be running
|
788 | 425f0f54 | Iustin Pop | def _check_ip(): |
789 | a744b676 | Manuel Franceschini | if netutils.TcpPing(master_ip, constants.DEFAULT_NODED_PORT):
|
790 | 425f0f54 | Iustin Pop | raise utils.RetryAgain()
|
791 | 425f0f54 | Iustin Pop | |
792 | 425f0f54 | Iustin Pop | try:
|
793 | 425f0f54 | Iustin Pop | utils.Retry(_check_ip, (1, 1.5, 5), total_timeout) |
794 | 425f0f54 | Iustin Pop | except utils.RetryTimeout:
|
795 | 425f0f54 | Iustin Pop | logging.warning("The master IP is still reachable after %s seconds,"
|
796 | 425f0f54 | Iustin Pop | " continuing but activating the master on the current"
|
797 | 425f0f54 | Iustin Pop | " node will probably fail", total_timeout)
|
798 | b1b6ea87 | Iustin Pop | |
799 | ff699aa9 | Michael Hanselmann | if jstore.CheckDrainFlag():
|
800 | ff699aa9 | Michael Hanselmann | logging.info("Undraining job queue")
|
801 | ff699aa9 | Michael Hanselmann | jstore.SetDrainFlag(False)
|
802 | ff699aa9 | Michael Hanselmann | |
803 | 21004460 | Iustin Pop | logging.info("Starting the master daemons on the new master")
|
804 | d5927e48 | Iustin Pop | |
805 | db04ce5d | Michael Hanselmann | result = rpc.BootstrapRunner().call_node_start_master_daemons(new_master, |
806 | db04ce5d | Michael Hanselmann | no_voting) |
807 | 3cebe102 | Michael Hanselmann | msg = result.fail_msg |
808 | b726aff0 | Iustin Pop | if msg:
|
809 | d5927e48 | Iustin Pop | logging.error("Could not start the master role on the new master"
|
810 | b726aff0 | Iustin Pop | " %s, please check: %s", new_master, msg)
|
811 | b1b6ea87 | Iustin Pop | rcode = 1
|
812 | b1b6ea87 | Iustin Pop | |
813 | 21004460 | Iustin Pop | logging.info("Master failed over from %s to %s", old_master, new_master)
|
814 | b1b6ea87 | Iustin Pop | return rcode
|
815 | d7cdb55d | Iustin Pop | |
816 | d7cdb55d | Iustin Pop | |
817 | 8eb148ae | Iustin Pop | def GetMaster(): |
818 | 8eb148ae | Iustin Pop | """Returns the current master node.
|
819 | 8eb148ae | Iustin Pop |
|
820 | 8eb148ae | Iustin Pop | This is a separate function in bootstrap since it's needed by
|
821 | 8eb148ae | Iustin Pop | gnt-cluster, and instead of importing directly ssconf, it's better
|
822 | 8eb148ae | Iustin Pop | to abstract it in bootstrap, where we do use ssconf in other
|
823 | 8eb148ae | Iustin Pop | functions too.
|
824 | 8eb148ae | Iustin Pop |
|
825 | 8eb148ae | Iustin Pop | """
|
826 | 8eb148ae | Iustin Pop | sstore = ssconf.SimpleStore() |
827 | 8eb148ae | Iustin Pop | |
828 | 8eb148ae | Iustin Pop | old_master, _ = ssconf.GetMasterAndMyself(sstore) |
829 | 8eb148ae | Iustin Pop | |
830 | 8eb148ae | Iustin Pop | return old_master
|
831 | 8eb148ae | Iustin Pop | |
832 | 8eb148ae | Iustin Pop | |
833 | d7cdb55d | Iustin Pop | def GatherMasterVotes(node_list): |
834 | d7cdb55d | Iustin Pop | """Check the agreement on who is the master.
|
835 | d7cdb55d | Iustin Pop |
|
836 | d7cdb55d | Iustin Pop | This function will return a list of (node, number of votes), ordered
|
837 | d7cdb55d | Iustin Pop | by the number of votes. Errors will be denoted by the key 'None'.
|
838 | d7cdb55d | Iustin Pop |
|
839 | d7cdb55d | Iustin Pop | Note that the sum of votes is the number of nodes this machine
|
840 | d7cdb55d | Iustin Pop | knows, whereas the number of entries in the list could be different
|
841 | d7cdb55d | Iustin Pop | (if some nodes vote for another master).
|
842 | d7cdb55d | Iustin Pop |
|
843 | d7cdb55d | Iustin Pop | We remove ourselves from the list since we know that (bugs aside)
|
844 | d7cdb55d | Iustin Pop | since we use the same source for configuration information for both
|
845 | d7cdb55d | Iustin Pop | backend and boostrap, we'll always vote for ourselves.
|
846 | d7cdb55d | Iustin Pop |
|
847 | d7cdb55d | Iustin Pop | @type node_list: list
|
848 | d7cdb55d | Iustin Pop | @param node_list: the list of nodes to query for master info; the current
|
849 | 5bbd3f7f | Michael Hanselmann | node will be removed if it is in the list
|
850 | d7cdb55d | Iustin Pop | @rtype: list
|
851 | d7cdb55d | Iustin Pop | @return: list of (node, votes)
|
852 | d7cdb55d | Iustin Pop |
|
853 | d7cdb55d | Iustin Pop | """
|
854 | b705c7a6 | Manuel Franceschini | myself = netutils.Hostname.GetSysName() |
855 | d7cdb55d | Iustin Pop | try:
|
856 | d7cdb55d | Iustin Pop | node_list.remove(myself) |
857 | d7cdb55d | Iustin Pop | except ValueError: |
858 | d7cdb55d | Iustin Pop | pass
|
859 | d7cdb55d | Iustin Pop | if not node_list: |
860 | d7cdb55d | Iustin Pop | # no nodes left (eventually after removing myself)
|
861 | d7cdb55d | Iustin Pop | return []
|
862 | db04ce5d | Michael Hanselmann | results = rpc.BootstrapRunner().call_master_info(node_list) |
863 | d7cdb55d | Iustin Pop | if not isinstance(results, dict): |
864 | d7cdb55d | Iustin Pop | # this should not happen (unless internal error in rpc)
|
865 | d7cdb55d | Iustin Pop | logging.critical("Can't complete rpc call, aborting master startup")
|
866 | d7cdb55d | Iustin Pop | return [(None, len(node_list))] |
867 | d7cdb55d | Iustin Pop | votes = {} |
868 | d7cdb55d | Iustin Pop | for node in results: |
869 | 781de953 | Iustin Pop | nres = results[node] |
870 | 2a52a064 | Iustin Pop | data = nres.payload |
871 | 3cebe102 | Michael Hanselmann | msg = nres.fail_msg |
872 | 2a52a064 | Iustin Pop | fail = False
|
873 | 2a52a064 | Iustin Pop | if msg:
|
874 | 2a52a064 | Iustin Pop | logging.warning("Error contacting node %s: %s", node, msg)
|
875 | 2a52a064 | Iustin Pop | fail = True
|
876 | 909b3a0e | Andrea Spadaccini | # for now we accept both length 3, 4 and 5 (data[3] is primary ip version
|
877 | 909b3a0e | Andrea Spadaccini | # and data[4] is the master netmask)
|
878 | 2a52a064 | Iustin Pop | elif not isinstance(data, (tuple, list)) or len(data) < 3: |
879 | 2a52a064 | Iustin Pop | logging.warning("Invalid data received from node %s: %s", node, data)
|
880 | 2a52a064 | Iustin Pop | fail = True
|
881 | 2a52a064 | Iustin Pop | if fail:
|
882 | d7cdb55d | Iustin Pop | if None not in votes: |
883 | d7cdb55d | Iustin Pop | votes[None] = 0 |
884 | d7cdb55d | Iustin Pop | votes[None] += 1 |
885 | d7cdb55d | Iustin Pop | continue
|
886 | 781de953 | Iustin Pop | master_node = data[2]
|
887 | d7cdb55d | Iustin Pop | if master_node not in votes: |
888 | d7cdb55d | Iustin Pop | votes[master_node] = 0
|
889 | d7cdb55d | Iustin Pop | votes[master_node] += 1
|
890 | d7cdb55d | Iustin Pop | |
891 | d7cdb55d | Iustin Pop | vote_list = [v for v in votes.items()] |
892 | d7cdb55d | Iustin Pop | # sort first on number of votes then on name, since we want None
|
893 | d7cdb55d | Iustin Pop | # sorted later if we have the half of the nodes not responding, and
|
894 | d7cdb55d | Iustin Pop | # half voting all for the same master
|
895 | d7cdb55d | Iustin Pop | vote_list.sort(key=lambda x: (x[1], x[0]), reverse=True) |
896 | d7cdb55d | Iustin Pop | |
897 | d7cdb55d | Iustin Pop | return vote_list |