blob: e65a52fa985346cd7efd28d4cd45a264af01f3d4 [file] [log] [blame]
Aviv Keshet07f16242013-10-10 07:54:19 -07001import os, time, socket, shutil, glob, logging, traceback, tempfile
Simran Basi3b858a22015-03-17 16:23:24 -07002from multiprocessing import Lock
Aviv Keshet53a216a2013-08-27 13:58:46 -07003from autotest_lib.client.common_lib import autotemp, error
jadmanski31c49b72008-10-27 20:44:48 +00004from autotest_lib.server import utils, autotest
mblighe8b93af2009-01-30 00:45:53 +00005from autotest_lib.server.hosts import remote
mblighefccc1b2010-01-11 19:08:42 +00006from autotest_lib.client.common_lib.global_config import global_config
jadmanskica7da372008-10-21 16:26:52 +00007
Aviv Keshet53a216a2013-08-27 13:58:46 -07008# pylint: disable-msg=C0111
jadmanskica7da372008-10-21 16:26:52 +00009
mblighb86bfa12010-02-12 20:22:21 +000010get_value = global_config.get_config_value
11enable_master_ssh = get_value('AUTOSERV', 'enable_master_ssh', type=bool,
12 default=False)
mblighefccc1b2010-01-11 19:08:42 +000013
14
Fang Deng96667ca2013-08-01 17:46:18 -070015class AbstractSSHHost(remote.RemoteHost):
mblighbc9402b2009-12-29 01:15:34 +000016 """
17 This class represents a generic implementation of most of the
jadmanskica7da372008-10-21 16:26:52 +000018 framework necessary for controlling a host via ssh. It implements
19 almost all of the abstract Host methods, except for the core
mblighbc9402b2009-12-29 01:15:34 +000020 Host.run method.
21 """
jadmanskica7da372008-10-21 16:26:52 +000022
jadmanskif6562912008-10-21 17:59:01 +000023 def _initialize(self, hostname, user="root", port=22, password="",
24 *args, **dargs):
25 super(AbstractSSHHost, self)._initialize(hostname=hostname,
26 *args, **dargs)
Dan Shic07b8932014-12-11 15:22:30 -080027 # IP address is retrieved only on demand. Otherwise the host
28 # initialization will fail for host is not online.
29 self._ip = None
jadmanskica7da372008-10-21 16:26:52 +000030 self.user = user
31 self.port = port
32 self.password = password
showard6eafb492010-01-15 20:29:06 +000033 self._use_rsync = None
Fang Deng3af66202013-08-16 15:19:25 -070034 self.known_hosts_file = tempfile.mkstemp()[1]
jadmanskica7da372008-10-21 16:26:52 +000035
mblighefccc1b2010-01-11 19:08:42 +000036 """
37 Master SSH connection background job, socket temp directory and socket
38 control path option. If master-SSH is enabled, these fields will be
39 initialized by start_master_ssh when a new SSH connection is initiated.
40 """
41 self.master_ssh_job = None
42 self.master_ssh_tempdir = None
43 self.master_ssh_option = ''
44
Simran Basi3b858a22015-03-17 16:23:24 -070045 # Create a Lock to protect against race conditions.
46 self._lock = Lock()
47
showard6eafb492010-01-15 20:29:06 +000048
Dan Shic07b8932014-12-11 15:22:30 -080049 @property
50 def ip(self):
51 """@return IP address of the host.
52 """
53 if not self._ip:
54 self._ip = socket.getaddrinfo(self.hostname, None)[0][4][0]
55 return self._ip
56
57
Fang Deng96667ca2013-08-01 17:46:18 -070058 def make_ssh_command(self, user="root", port=22, opts='',
59 hosts_file='/dev/null',
60 connect_timeout=30, alive_interval=300):
61 base_command = ("/usr/bin/ssh -a -x %s -o StrictHostKeyChecking=no "
62 "-o UserKnownHostsFile=%s -o BatchMode=yes "
63 "-o ConnectTimeout=%d -o ServerAliveInterval=%d "
64 "-l %s -p %d")
65 assert isinstance(connect_timeout, (int, long))
66 assert connect_timeout > 0 # can't disable the timeout
67 return base_command % (opts, hosts_file, connect_timeout,
68 alive_interval, user, port)
69
70
showard6eafb492010-01-15 20:29:06 +000071 def use_rsync(self):
72 if self._use_rsync is not None:
73 return self._use_rsync
74
mblighc9892c02010-01-06 19:02:16 +000075 # Check if rsync is available on the remote host. If it's not,
76 # don't try to use it for any future file transfers.
showard6eafb492010-01-15 20:29:06 +000077 self._use_rsync = self._check_rsync()
78 if not self._use_rsync:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -070079 logging.warning("rsync not available on remote host %s -- disabled",
mblighc9892c02010-01-06 19:02:16 +000080 self.hostname)
Eric Lie0493a42010-11-15 13:05:43 -080081 return self._use_rsync
mblighc9892c02010-01-06 19:02:16 +000082
83
84 def _check_rsync(self):
85 """
86 Check if rsync is available on the remote host.
87 """
88 try:
89 self.run("rsync --version", stdout_tee=None, stderr_tee=None)
90 except error.AutoservRunError:
91 return False
92 return True
93
jadmanskica7da372008-10-21 16:26:52 +000094
showard56176ec2009-10-28 19:52:30 +000095 def _encode_remote_paths(self, paths, escape=True):
mblighbc9402b2009-12-29 01:15:34 +000096 """
97 Given a list of file paths, encodes it as a single remote path, in
98 the style used by rsync and scp.
99 """
showard56176ec2009-10-28 19:52:30 +0000100 if escape:
101 paths = [utils.scp_remote_escape(path) for path in paths]
102 return '%s@%s:"%s"' % (self.user, self.hostname, " ".join(paths))
jadmanskica7da372008-10-21 16:26:52 +0000103
jadmanskica7da372008-10-21 16:26:52 +0000104
mbligh45561782009-05-11 21:14:34 +0000105 def _make_rsync_cmd(self, sources, dest, delete_dest, preserve_symlinks):
mblighbc9402b2009-12-29 01:15:34 +0000106 """
107 Given a list of source paths and a destination path, produces the
jadmanskid7b79ed2009-01-07 17:19:48 +0000108 appropriate rsync command for copying them. Remote paths must be
mblighbc9402b2009-12-29 01:15:34 +0000109 pre-encoded.
110 """
Fang Deng96667ca2013-08-01 17:46:18 -0700111 ssh_cmd = self.make_ssh_command(user=self.user, port=self.port,
112 opts=self.master_ssh_option,
113 hosts_file=self.known_hosts_file)
jadmanskid7b79ed2009-01-07 17:19:48 +0000114 if delete_dest:
115 delete_flag = "--delete"
116 else:
117 delete_flag = ""
mbligh45561782009-05-11 21:14:34 +0000118 if preserve_symlinks:
119 symlink_flag = ""
120 else:
121 symlink_flag = "-L"
Dan Shi06d7fbf2014-02-12 12:34:41 -0800122 command = ("rsync %s %s --timeout=1800 --rsh='%s' -az --no-o --no-g "
David Hendricksb8904182014-06-02 15:22:49 -0700123 "%s \"%s\"")
mbligh45561782009-05-11 21:14:34 +0000124 return command % (symlink_flag, delete_flag, ssh_cmd,
David Hendricksb8904182014-06-02 15:22:49 -0700125 " ".join(['"%s"' % p for p in sources]), dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000126
127
Eric Li861b2d52011-02-04 14:50:35 -0800128 def _make_ssh_cmd(self, cmd):
129 """
130 Create a base ssh command string for the host which can be used
131 to run commands directly on the machine
132 """
Fang Deng96667ca2013-08-01 17:46:18 -0700133 base_cmd = self.make_ssh_command(user=self.user, port=self.port,
134 opts=self.master_ssh_option,
135 hosts_file=self.known_hosts_file)
Eric Li861b2d52011-02-04 14:50:35 -0800136
137 return '%s %s "%s"' % (base_cmd, self.hostname, utils.sh_escape(cmd))
138
jadmanskid7b79ed2009-01-07 17:19:48 +0000139 def _make_scp_cmd(self, sources, dest):
mblighbc9402b2009-12-29 01:15:34 +0000140 """
141 Given a list of source paths and a destination path, produces the
jadmanskid7b79ed2009-01-07 17:19:48 +0000142 appropriate scp command for encoding it. Remote paths must be
mblighbc9402b2009-12-29 01:15:34 +0000143 pre-encoded.
144 """
mblighc0649d62010-01-15 18:15:58 +0000145 command = ("scp -rq %s -o StrictHostKeyChecking=no "
lmraf676f32010-02-04 03:36:26 +0000146 "-o UserKnownHostsFile=%s -P %d %s '%s'")
Fang Deng3af66202013-08-16 15:19:25 -0700147 return command % (self.master_ssh_option, self.known_hosts_file,
mblighefccc1b2010-01-11 19:08:42 +0000148 self.port, " ".join(sources), dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000149
150
151 def _make_rsync_compatible_globs(self, path, is_local):
mblighbc9402b2009-12-29 01:15:34 +0000152 """
153 Given an rsync-style path, returns a list of globbed paths
jadmanskid7b79ed2009-01-07 17:19:48 +0000154 that will hopefully provide equivalent behaviour for scp. Does not
155 support the full range of rsync pattern matching behaviour, only that
156 exposed in the get/send_file interface (trailing slashes).
157
158 The is_local param is flag indicating if the paths should be
mblighbc9402b2009-12-29 01:15:34 +0000159 interpreted as local or remote paths.
160 """
jadmanskid7b79ed2009-01-07 17:19:48 +0000161
162 # non-trailing slash paths should just work
163 if len(path) == 0 or path[-1] != "/":
164 return [path]
165
166 # make a function to test if a pattern matches any files
167 if is_local:
showard56176ec2009-10-28 19:52:30 +0000168 def glob_matches_files(path, pattern):
169 return len(glob.glob(path + pattern)) > 0
jadmanskid7b79ed2009-01-07 17:19:48 +0000170 else:
showard56176ec2009-10-28 19:52:30 +0000171 def glob_matches_files(path, pattern):
172 result = self.run("ls \"%s\"%s" % (utils.sh_escape(path),
173 pattern),
174 stdout_tee=None, ignore_status=True)
jadmanskid7b79ed2009-01-07 17:19:48 +0000175 return result.exit_status == 0
176
177 # take a set of globs that cover all files, and see which are needed
178 patterns = ["*", ".[!.]*"]
showard56176ec2009-10-28 19:52:30 +0000179 patterns = [p for p in patterns if glob_matches_files(path, p)]
jadmanskid7b79ed2009-01-07 17:19:48 +0000180
181 # convert them into a set of paths suitable for the commandline
jadmanskid7b79ed2009-01-07 17:19:48 +0000182 if is_local:
showard56176ec2009-10-28 19:52:30 +0000183 return ["\"%s\"%s" % (utils.sh_escape(path), pattern)
184 for pattern in patterns]
jadmanskid7b79ed2009-01-07 17:19:48 +0000185 else:
showard56176ec2009-10-28 19:52:30 +0000186 return [utils.scp_remote_escape(path) + pattern
187 for pattern in patterns]
jadmanskid7b79ed2009-01-07 17:19:48 +0000188
189
190 def _make_rsync_compatible_source(self, source, is_local):
mblighbc9402b2009-12-29 01:15:34 +0000191 """
192 Applies the same logic as _make_rsync_compatible_globs, but
jadmanskid7b79ed2009-01-07 17:19:48 +0000193 applies it to an entire list of sources, producing a new list of
mblighbc9402b2009-12-29 01:15:34 +0000194 sources, properly quoted.
195 """
jadmanskid7b79ed2009-01-07 17:19:48 +0000196 return sum((self._make_rsync_compatible_globs(path, is_local)
197 for path in source), [])
jadmanskica7da372008-10-21 16:26:52 +0000198
199
mblighfeac0102009-04-28 18:31:12 +0000200 def _set_umask_perms(self, dest):
mblighbc9402b2009-12-29 01:15:34 +0000201 """
202 Given a destination file/dir (recursively) set the permissions on
203 all the files and directories to the max allowed by running umask.
204 """
mblighfeac0102009-04-28 18:31:12 +0000205
206 # now this looks strange but I haven't found a way in Python to _just_
207 # get the umask, apparently the only option is to try to set it
208 umask = os.umask(0)
209 os.umask(umask)
210
211 max_privs = 0777 & ~umask
212
213 def set_file_privs(filename):
Chris Masone567d0d92011-12-19 09:38:30 -0800214 """Sets mode of |filename|. Assumes |filename| exists."""
215 file_stat = os.stat(filename)
mblighfeac0102009-04-28 18:31:12 +0000216
217 file_privs = max_privs
218 # if the original file permissions do not have at least one
219 # executable bit then do not set it anywhere
220 if not file_stat.st_mode & 0111:
221 file_privs &= ~0111
222
223 os.chmod(filename, file_privs)
224
225 # try a bottom-up walk so changes on directory permissions won't cut
226 # our access to the files/directories inside it
227 for root, dirs, files in os.walk(dest, topdown=False):
228 # when setting the privileges we emulate the chmod "X" behaviour
229 # that sets to execute only if it is a directory or any of the
230 # owner/group/other already has execute right
231 for dirname in dirs:
232 os.chmod(os.path.join(root, dirname), max_privs)
233
Chris Masone567d0d92011-12-19 09:38:30 -0800234 # Filter out broken symlinks as we go.
235 for filename in filter(os.path.exists, files):
mblighfeac0102009-04-28 18:31:12 +0000236 set_file_privs(os.path.join(root, filename))
237
238
239 # now set privs for the dest itself
240 if os.path.isdir(dest):
241 os.chmod(dest, max_privs)
242 else:
243 set_file_privs(dest)
244
245
mbligh45561782009-05-11 21:14:34 +0000246 def get_file(self, source, dest, delete_dest=False, preserve_perm=True,
247 preserve_symlinks=False):
jadmanskica7da372008-10-21 16:26:52 +0000248 """
249 Copy files from the remote host to a local path.
250
251 Directories will be copied recursively.
252 If a source component is a directory with a trailing slash,
253 the content of the directory will be copied, otherwise, the
254 directory itself and its content will be copied. This
255 behavior is similar to that of the program 'rsync'.
256
257 Args:
258 source: either
259 1) a single file or directory, as a string
260 2) a list of one or more (possibly mixed)
261 files or directories
262 dest: a file or a directory (if source contains a
263 directory or more than one element, you must
264 supply a directory dest)
mbligh89e258d2008-10-24 13:58:08 +0000265 delete_dest: if this is true, the command will also clear
266 out any old files at dest that are not in the
267 source
mblighfeac0102009-04-28 18:31:12 +0000268 preserve_perm: tells get_file() to try to preserve the sources
269 permissions on files and dirs
mbligh45561782009-05-11 21:14:34 +0000270 preserve_symlinks: try to preserve symlinks instead of
271 transforming them into files/dirs on copy
jadmanskica7da372008-10-21 16:26:52 +0000272
273 Raises:
274 AutoservRunError: the scp command failed
275 """
Simran Basi882f15b2013-10-29 14:59:34 -0700276 logging.debug('get_file. source: %s, dest: %s, delete_dest: %s,'
277 'preserve_perm: %s, preserve_symlinks:%s', source, dest,
278 delete_dest, preserve_perm, preserve_symlinks)
mblighefccc1b2010-01-11 19:08:42 +0000279 # Start a master SSH connection if necessary.
280 self.start_master_ssh()
281
jadmanskica7da372008-10-21 16:26:52 +0000282 if isinstance(source, basestring):
283 source = [source]
jadmanskid7b79ed2009-01-07 17:19:48 +0000284 dest = os.path.abspath(dest)
jadmanskica7da372008-10-21 16:26:52 +0000285
mblighc9892c02010-01-06 19:02:16 +0000286 # If rsync is disabled or fails, try scp.
showard6eafb492010-01-15 20:29:06 +0000287 try_scp = True
288 if self.use_rsync():
Simran Basi882f15b2013-10-29 14:59:34 -0700289 logging.debug('Using Rsync.')
mblighc9892c02010-01-06 19:02:16 +0000290 try:
291 remote_source = self._encode_remote_paths(source)
292 local_dest = utils.sh_escape(dest)
293 rsync = self._make_rsync_cmd([remote_source], local_dest,
294 delete_dest, preserve_symlinks)
295 utils.run(rsync)
showard6eafb492010-01-15 20:29:06 +0000296 try_scp = False
mblighc9892c02010-01-06 19:02:16 +0000297 except error.CmdError, e:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -0700298 logging.warning("trying scp, rsync failed: %s", e)
mblighc9892c02010-01-06 19:02:16 +0000299
300 if try_scp:
Simran Basi882f15b2013-10-29 14:59:34 -0700301 logging.debug('Trying scp.')
jadmanskid7b79ed2009-01-07 17:19:48 +0000302 # scp has no equivalent to --delete, just drop the entire dest dir
303 if delete_dest and os.path.isdir(dest):
304 shutil.rmtree(dest)
305 os.mkdir(dest)
jadmanskica7da372008-10-21 16:26:52 +0000306
jadmanskid7b79ed2009-01-07 17:19:48 +0000307 remote_source = self._make_rsync_compatible_source(source, False)
308 if remote_source:
showard56176ec2009-10-28 19:52:30 +0000309 # _make_rsync_compatible_source() already did the escaping
310 remote_source = self._encode_remote_paths(remote_source,
311 escape=False)
jadmanskid7b79ed2009-01-07 17:19:48 +0000312 local_dest = utils.sh_escape(dest)
jadmanski2583a432009-02-10 23:59:11 +0000313 scp = self._make_scp_cmd([remote_source], local_dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000314 try:
315 utils.run(scp)
316 except error.CmdError, e:
Simran Basi882f15b2013-10-29 14:59:34 -0700317 logging.debug('scp failed: %s', e)
jadmanskid7b79ed2009-01-07 17:19:48 +0000318 raise error.AutoservRunError(e.args[0], e.args[1])
jadmanskica7da372008-10-21 16:26:52 +0000319
mblighfeac0102009-04-28 18:31:12 +0000320 if not preserve_perm:
321 # we have no way to tell scp to not try to preserve the
322 # permissions so set them after copy instead.
323 # for rsync we could use "--no-p --chmod=ugo=rwX" but those
324 # options are only in very recent rsync versions
325 self._set_umask_perms(dest)
326
jadmanskica7da372008-10-21 16:26:52 +0000327
mbligh45561782009-05-11 21:14:34 +0000328 def send_file(self, source, dest, delete_dest=False,
329 preserve_symlinks=False):
jadmanskica7da372008-10-21 16:26:52 +0000330 """
331 Copy files from a local path to the remote host.
332
333 Directories will be copied recursively.
334 If a source component is a directory with a trailing slash,
335 the content of the directory will be copied, otherwise, the
336 directory itself and its content will be copied. This
337 behavior is similar to that of the program 'rsync'.
338
339 Args:
340 source: either
341 1) a single file or directory, as a string
342 2) a list of one or more (possibly mixed)
343 files or directories
344 dest: a file or a directory (if source contains a
345 directory or more than one element, you must
346 supply a directory dest)
mbligh89e258d2008-10-24 13:58:08 +0000347 delete_dest: if this is true, the command will also clear
348 out any old files at dest that are not in the
349 source
mbligh45561782009-05-11 21:14:34 +0000350 preserve_symlinks: controls if symlinks on the source will be
351 copied as such on the destination or transformed into the
352 referenced file/directory
jadmanskica7da372008-10-21 16:26:52 +0000353
354 Raises:
355 AutoservRunError: the scp command failed
356 """
Simran Basi882f15b2013-10-29 14:59:34 -0700357 logging.debug('send_file. source: %s, dest: %s, delete_dest: %s,'
358 'preserve_symlinks:%s', source, dest,
359 delete_dest, preserve_symlinks)
mblighefccc1b2010-01-11 19:08:42 +0000360 # Start a master SSH connection if necessary.
361 self.start_master_ssh()
362
jadmanskica7da372008-10-21 16:26:52 +0000363 if isinstance(source, basestring):
364 source = [source]
jadmanski2583a432009-02-10 23:59:11 +0000365 remote_dest = self._encode_remote_paths([dest])
jadmanskica7da372008-10-21 16:26:52 +0000366
mukesh agrawal0d3616c2015-07-17 15:47:36 -0700367 local_sources = [utils.sh_escape(path) for path in source]
368 if not local_sources:
369 raise error.TestError('source |%s| yielded an empty list' % (
370 source))
371 if any([local_source.find('\x00') != -1 for
372 local_source in local_sources]):
373 raise error.TestError('one or more sources include NUL char')
374
mblighc9892c02010-01-06 19:02:16 +0000375 # If rsync is disabled or fails, try scp.
showard6eafb492010-01-15 20:29:06 +0000376 try_scp = True
377 if self.use_rsync():
Simran Basi882f15b2013-10-29 14:59:34 -0700378 logging.debug('Using Rsync.')
mblighc9892c02010-01-06 19:02:16 +0000379 try:
mblighc9892c02010-01-06 19:02:16 +0000380 rsync = self._make_rsync_cmd(local_sources, remote_dest,
381 delete_dest, preserve_symlinks)
382 utils.run(rsync)
showard6eafb492010-01-15 20:29:06 +0000383 try_scp = False
mblighc9892c02010-01-06 19:02:16 +0000384 except error.CmdError, e:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -0700385 logging.warning("trying scp, rsync failed: %s", e)
mblighc9892c02010-01-06 19:02:16 +0000386
387 if try_scp:
Simran Basi882f15b2013-10-29 14:59:34 -0700388 logging.debug('Trying scp.')
jadmanskid7b79ed2009-01-07 17:19:48 +0000389 # scp has no equivalent to --delete, just drop the entire dest dir
390 if delete_dest:
showard27160152009-07-15 14:28:42 +0000391 is_dir = self.run("ls -d %s/" % dest,
jadmanskid7b79ed2009-01-07 17:19:48 +0000392 ignore_status=True).exit_status == 0
393 if is_dir:
394 cmd = "rm -rf %s && mkdir %s"
mbligh5a0ca532009-08-03 16:44:34 +0000395 cmd %= (dest, dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000396 self.run(cmd)
jadmanskica7da372008-10-21 16:26:52 +0000397
jadmanski2583a432009-02-10 23:59:11 +0000398 local_sources = self._make_rsync_compatible_source(source, True)
399 if local_sources:
400 scp = self._make_scp_cmd(local_sources, remote_dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000401 try:
402 utils.run(scp)
403 except error.CmdError, e:
Simran Basi882f15b2013-10-29 14:59:34 -0700404 logging.debug('scp failed: %s', e)
jadmanskid7b79ed2009-01-07 17:19:48 +0000405 raise error.AutoservRunError(e.args[0], e.args[1])
mukesh agrawal0d3616c2015-07-17 15:47:36 -0700406 else:
407 logging.debug('skipping scp for empty source list')
jadmanskid7b79ed2009-01-07 17:19:48 +0000408
jadmanskica7da372008-10-21 16:26:52 +0000409
beeps46dadc92013-11-07 14:07:10 -0800410 def ssh_ping(self, timeout=60, base_cmd='true'):
beepsadd66d32013-03-04 17:21:51 -0800411 """
412 Pings remote host via ssh.
413
414 @param timeout: Time in seconds before giving up.
415 Defaults to 60 seconds.
beeps46dadc92013-11-07 14:07:10 -0800416 @param base_cmd: The base command to run with the ssh ping.
417 Defaults to true.
beepsadd66d32013-03-04 17:21:51 -0800418 @raise AutoservSSHTimeout: If the ssh ping times out.
419 @raise AutoservSshPermissionDeniedError: If ssh ping fails due to
420 permissions.
421 @raise AutoservSshPingHostError: For other AutoservRunErrors.
422 """
jadmanskica7da372008-10-21 16:26:52 +0000423 try:
beeps46dadc92013-11-07 14:07:10 -0800424 self.run(base_cmd, timeout=timeout, connect_timeout=timeout)
jadmanskica7da372008-10-21 16:26:52 +0000425 except error.AutoservSSHTimeout:
mblighd0e94982009-07-11 00:15:18 +0000426 msg = "Host (ssh) verify timed out (timeout = %d)" % timeout
jadmanskica7da372008-10-21 16:26:52 +0000427 raise error.AutoservSSHTimeout(msg)
mbligh9d738d62009-03-09 21:17:10 +0000428 except error.AutoservSshPermissionDeniedError:
429 #let AutoservSshPermissionDeniedError be visible to the callers
430 raise
jadmanskica7da372008-10-21 16:26:52 +0000431 except error.AutoservRunError, e:
mblighc971c5f2009-06-08 16:48:54 +0000432 # convert the generic AutoservRunError into something more
433 # specific for this context
434 raise error.AutoservSshPingHostError(e.description + '\n' +
435 repr(e.result_obj))
jadmanskica7da372008-10-21 16:26:52 +0000436
437
beeps46dadc92013-11-07 14:07:10 -0800438 def is_up(self, timeout=60, base_cmd='true'):
jadmanskica7da372008-10-21 16:26:52 +0000439 """
beeps46dadc92013-11-07 14:07:10 -0800440 Check if the remote host is up by ssh-ing and running a base command.
jadmanskica7da372008-10-21 16:26:52 +0000441
beepsadd66d32013-03-04 17:21:51 -0800442 @param timeout: timeout in seconds.
beeps46dadc92013-11-07 14:07:10 -0800443 @param base_cmd: a base command to run with ssh. The default is 'true'.
beepsadd66d32013-03-04 17:21:51 -0800444 @returns True if the remote host is up before the timeout expires,
445 False otherwise.
jadmanskica7da372008-10-21 16:26:52 +0000446 """
447 try:
beeps46dadc92013-11-07 14:07:10 -0800448 self.ssh_ping(timeout=timeout, base_cmd=base_cmd)
jadmanskica7da372008-10-21 16:26:52 +0000449 except error.AutoservError:
450 return False
451 else:
452 return True
453
454
455 def wait_up(self, timeout=None):
456 """
457 Wait until the remote host is up or the timeout expires.
458
459 In fact, it will wait until an ssh connection to the remote
460 host can be established, and getty is running.
461
jadmanskic0354912010-01-12 15:57:29 +0000462 @param timeout time limit in seconds before returning even
463 if the host is not up.
jadmanskica7da372008-10-21 16:26:52 +0000464
beepsadd66d32013-03-04 17:21:51 -0800465 @returns True if the host was found to be up before the timeout expires,
466 False otherwise
jadmanskica7da372008-10-21 16:26:52 +0000467 """
468 if timeout:
beeps46dadc92013-11-07 14:07:10 -0800469 current_time = int(time.time())
470 end_time = current_time + timeout
jadmanskica7da372008-10-21 16:26:52 +0000471
beepsadd66d32013-03-04 17:21:51 -0800472 while not timeout or current_time < end_time:
473 if self.is_up(timeout=end_time - current_time):
jadmanskica7da372008-10-21 16:26:52 +0000474 try:
475 if self.are_wait_up_processes_up():
jadmanski7ebac3d2010-06-17 16:06:31 +0000476 logging.debug('Host %s is now up', self.hostname)
jadmanskica7da372008-10-21 16:26:52 +0000477 return True
478 except error.AutoservError:
479 pass
480 time.sleep(1)
beeps46dadc92013-11-07 14:07:10 -0800481 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000482
jadmanski7ebac3d2010-06-17 16:06:31 +0000483 logging.debug('Host %s is still down after waiting %d seconds',
484 self.hostname, int(timeout + time.time() - end_time))
jadmanskica7da372008-10-21 16:26:52 +0000485 return False
486
487
jadmanskic0354912010-01-12 15:57:29 +0000488 def wait_down(self, timeout=None, warning_timer=None, old_boot_id=None):
jadmanskica7da372008-10-21 16:26:52 +0000489 """
490 Wait until the remote host is down or the timeout expires.
491
jadmanskic0354912010-01-12 15:57:29 +0000492 If old_boot_id is provided, this will wait until either the machine
493 is unpingable or self.get_boot_id() returns a value different from
494 old_boot_id. If the boot_id value has changed then the function
495 returns true under the assumption that the machine has shut down
496 and has now already come back up.
jadmanskica7da372008-10-21 16:26:52 +0000497
jadmanskic0354912010-01-12 15:57:29 +0000498 If old_boot_id is None then until the machine becomes unreachable the
499 method assumes the machine has not yet shut down.
jadmanskica7da372008-10-21 16:26:52 +0000500
beepsadd66d32013-03-04 17:21:51 -0800501 Based on this definition, the 4 possible permutations of timeout
502 and old_boot_id are:
503 1. timeout and old_boot_id: wait timeout seconds for either the
504 host to become unpingable, or the boot id
505 to change. In the latter case we've rebooted
506 and in the former case we've only shutdown,
507 but both cases return True.
508 2. only timeout: wait timeout seconds for the host to become unpingable.
509 If the host remains pingable throughout timeout seconds
510 we return False.
511 3. only old_boot_id: wait forever until either the host becomes
512 unpingable or the boot_id changes. Return true
513 when either of those conditions are met.
514 4. not timeout, not old_boot_id: wait forever till the host becomes
515 unpingable.
516
jadmanskic0354912010-01-12 15:57:29 +0000517 @param timeout Time limit in seconds before returning even
518 if the host is still up.
519 @param warning_timer Time limit in seconds that will generate
520 a warning if the host is not down yet.
521 @param old_boot_id A string containing the result of self.get_boot_id()
522 prior to the host being told to shut down. Can be None if this is
523 not available.
524
525 @returns True if the host was found to be down, False otherwise
jadmanskica7da372008-10-21 16:26:52 +0000526 """
mblighe5e3cf22010-05-27 23:33:14 +0000527 #TODO: there is currently no way to distinguish between knowing
528 #TODO: boot_id was unsupported and not knowing the boot_id.
beeps46dadc92013-11-07 14:07:10 -0800529 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000530 if timeout:
mbligh2ed998f2009-04-08 21:03:47 +0000531 end_time = current_time + timeout
jadmanskica7da372008-10-21 16:26:52 +0000532
mbligh2ed998f2009-04-08 21:03:47 +0000533 if warning_timer:
534 warn_time = current_time + warning_timer
535
jadmanskic0354912010-01-12 15:57:29 +0000536 if old_boot_id is not None:
537 logging.debug('Host %s pre-shutdown boot_id is %s',
538 self.hostname, old_boot_id)
539
beepsadd66d32013-03-04 17:21:51 -0800540 # Impose semi real-time deadline constraints, since some clients
541 # (eg: watchdog timer tests) expect strict checking of time elapsed.
542 # Each iteration of this loop is treated as though it atomically
543 # completes within current_time, this is needed because if we used
544 # inline time.time() calls instead then the following could happen:
545 #
546 # while not timeout or time.time() < end_time: [23 < 30]
547 # some code. [takes 10 secs]
548 # try:
549 # new_boot_id = self.get_boot_id(timeout=end_time - time.time())
550 # [30 - 33]
551 # The last step will lead to a return True, when in fact the machine
552 # went down at 32 seconds (>30). Hence we need to pass get_boot_id
553 # the same time that allowed us into that iteration of the loop.
mbligh2ed998f2009-04-08 21:03:47 +0000554 while not timeout or current_time < end_time:
jadmanskic0354912010-01-12 15:57:29 +0000555 try:
beeps46dadc92013-11-07 14:07:10 -0800556 new_boot_id = self.get_boot_id(timeout=end_time-current_time)
mblighdbc7e4a2010-01-15 20:34:20 +0000557 except error.AutoservError:
jadmanskic0354912010-01-12 15:57:29 +0000558 logging.debug('Host %s is now unreachable over ssh, is down',
559 self.hostname)
jadmanskica7da372008-10-21 16:26:52 +0000560 return True
jadmanskic0354912010-01-12 15:57:29 +0000561 else:
562 # if the machine is up but the boot_id value has changed from
563 # old boot id, then we can assume the machine has gone down
564 # and then already come back up
565 if old_boot_id is not None and old_boot_id != new_boot_id:
566 logging.debug('Host %s now has boot_id %s and so must '
567 'have rebooted', self.hostname, new_boot_id)
568 return True
mbligh2ed998f2009-04-08 21:03:47 +0000569
570 if warning_timer and current_time > warn_time:
Scott Zawalskic86fdeb2013-10-23 10:24:04 -0400571 self.record("INFO", None, "shutdown",
mbligh2ed998f2009-04-08 21:03:47 +0000572 "Shutdown took longer than %ds" % warning_timer)
573 # Print the warning only once.
574 warning_timer = None
mbligha4464402009-04-17 20:13:41 +0000575 # If a machine is stuck switching runlevels
576 # This may cause the machine to reboot.
577 self.run('kill -HUP 1', ignore_status=True)
mbligh2ed998f2009-04-08 21:03:47 +0000578
jadmanskica7da372008-10-21 16:26:52 +0000579 time.sleep(1)
beeps46dadc92013-11-07 14:07:10 -0800580 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000581
582 return False
jadmanskif6562912008-10-21 17:59:01 +0000583
mbligha0a27592009-01-24 01:41:36 +0000584
jadmanskif6562912008-10-21 17:59:01 +0000585 # tunable constants for the verify & repair code
mblighb86bfa12010-02-12 20:22:21 +0000586 AUTOTEST_GB_DISKSPACE_REQUIRED = get_value("SERVER",
587 "gb_diskspace_required",
Fang Deng6b05f5b2013-03-20 13:42:11 -0700588 type=float,
589 default=20.0)
mbligha0a27592009-01-24 01:41:36 +0000590
jadmanskif6562912008-10-21 17:59:01 +0000591
showardca572982009-09-18 21:20:01 +0000592 def verify_connectivity(self):
593 super(AbstractSSHHost, self).verify_connectivity()
jadmanskif6562912008-10-21 17:59:01 +0000594
showardb18134f2009-03-20 20:52:18 +0000595 logging.info('Pinging host ' + self.hostname)
jadmanskif6562912008-10-21 17:59:01 +0000596 self.ssh_ping()
mbligh2ba7ab02009-08-24 22:09:26 +0000597 logging.info("Host (ssh) %s is alive", self.hostname)
jadmanskif6562912008-10-21 17:59:01 +0000598
jadmanski80deb752009-01-21 17:14:16 +0000599 if self.is_shutting_down():
mblighc971c5f2009-06-08 16:48:54 +0000600 raise error.AutoservHostIsShuttingDownError("Host is shutting down")
jadmanski80deb752009-01-21 17:14:16 +0000601
mblighb49b5232009-02-12 21:54:49 +0000602
showardca572982009-09-18 21:20:01 +0000603 def verify_software(self):
604 super(AbstractSSHHost, self).verify_software()
jadmanskif6562912008-10-21 17:59:01 +0000605 try:
showardad812bf2009-10-20 23:49:56 +0000606 self.check_diskspace(autotest.Autotest.get_install_dir(self),
607 self.AUTOTEST_GB_DISKSPACE_REQUIRED)
jadmanskif6562912008-10-21 17:59:01 +0000608 except error.AutoservHostError:
609 raise # only want to raise if it's a space issue
showardad812bf2009-10-20 23:49:56 +0000610 except autotest.AutodirNotFoundError:
showardca572982009-09-18 21:20:01 +0000611 # autotest dir may not exist, etc. ignore
612 logging.debug('autodir space check exception, this is probably '
613 'safe to ignore\n' + traceback.format_exc())
mblighefccc1b2010-01-11 19:08:42 +0000614
615
616 def close(self):
617 super(AbstractSSHHost, self).close()
618 self._cleanup_master_ssh()
Fang Deng3af66202013-08-16 15:19:25 -0700619 os.remove(self.known_hosts_file)
mblighefccc1b2010-01-11 19:08:42 +0000620
621
622 def _cleanup_master_ssh(self):
623 """
624 Release all resources (process, temporary directory) used by an active
625 master SSH connection.
626 """
627 # If a master SSH connection is running, kill it.
628 if self.master_ssh_job is not None:
Aviv Keshet46250752013-08-27 15:52:06 -0700629 logging.debug('Nuking master_ssh_job.')
mblighefccc1b2010-01-11 19:08:42 +0000630 utils.nuke_subprocess(self.master_ssh_job.sp)
631 self.master_ssh_job = None
632
633 # Remove the temporary directory for the master SSH socket.
634 if self.master_ssh_tempdir is not None:
Aviv Keshet46250752013-08-27 15:52:06 -0700635 logging.debug('Cleaning master_ssh_tempdir.')
mblighefccc1b2010-01-11 19:08:42 +0000636 self.master_ssh_tempdir.clean()
637 self.master_ssh_tempdir = None
638 self.master_ssh_option = ''
639
640
Aviv Keshet0749a822013-10-17 09:53:26 -0700641 def start_master_ssh(self, timeout=5):
mblighefccc1b2010-01-11 19:08:42 +0000642 """
643 Called whenever a slave SSH connection needs to be initiated (e.g., by
644 run, rsync, scp). If master SSH support is enabled and a master SSH
645 connection is not active already, start a new one in the background.
646 Also, cleanup any zombie master SSH connections (e.g., dead due to
647 reboot).
Aviv Keshet0749a822013-10-17 09:53:26 -0700648
649 timeout: timeout in seconds (default 5) to wait for master ssh
650 connection to be established. If timeout is reached, a
651 warning message is logged, but no other action is taken.
mblighefccc1b2010-01-11 19:08:42 +0000652 """
653 if not enable_master_ssh:
654 return
655
Simran Basi3b858a22015-03-17 16:23:24 -0700656 # Multiple processes might try in parallel to clean up the old master
657 # ssh connection and create a new one, therefore use a lock to protect
658 # against race conditions.
659 with self._lock:
660 # If a previously started master SSH connection is not running
661 # anymore, it needs to be cleaned up and then restarted.
662 if self.master_ssh_job is not None:
663 socket_path = os.path.join(self.master_ssh_tempdir.name,
664 'socket')
665 if (not os.path.exists(socket_path) or
666 self.master_ssh_job.sp.poll() is not None):
667 logging.info("Master ssh connection to %s is down.",
668 self.hostname)
669 self._cleanup_master_ssh()
mblighefccc1b2010-01-11 19:08:42 +0000670
Simran Basi3b858a22015-03-17 16:23:24 -0700671 # Start a new master SSH connection.
672 if self.master_ssh_job is None:
673 # Create a shared socket in a temp location.
674 self.master_ssh_tempdir = autotemp.tempdir(
675 unique_id='ssh-master')
676 self.master_ssh_option = ("-o ControlPath=%s/socket" %
677 self.master_ssh_tempdir.name)
mblighefccc1b2010-01-11 19:08:42 +0000678
Simran Basi3b858a22015-03-17 16:23:24 -0700679 # Start the master SSH connection in the background.
680 master_cmd = self.ssh_command(
681 options="-N -o ControlMaster=yes")
682 logging.info("Starting master ssh connection '%s'", master_cmd)
683 self.master_ssh_job = utils.BgJob(master_cmd,
684 nickname='master-ssh',
685 no_pipes=True)
686 # To prevent a race between the the master ssh connection
687 # startup and its first attempted use, wait for socket file to
688 # exist before returning.
689 end_time = time.time() + timeout
690 socket_file_path = os.path.join(self.master_ssh_tempdir.name,
691 'socket')
692 while time.time() < end_time:
693 if os.path.exists(socket_file_path):
694 break
695 time.sleep(.2)
696 else:
697 logging.info('Timed out waiting for master-ssh connection '
698 'to be established.')
mbligh0a883702010-04-21 01:58:34 +0000699
700
701 def clear_known_hosts(self):
702 """Clears out the temporary ssh known_hosts file.
703
704 This is useful if the test SSHes to the machine, then reinstalls it,
705 then SSHes to it again. It can be called after the reinstall to
706 reduce the spam in the logs.
707 """
708 logging.info("Clearing known hosts for host '%s', file '%s'.",
Fang Deng3af66202013-08-16 15:19:25 -0700709 self.hostname, self.known_hosts_file)
mbligh0a883702010-04-21 01:58:34 +0000710 # Clear out the file by opening it for writing and then closing.
Fang Deng3af66202013-08-16 15:19:25 -0700711 fh = open(self.known_hosts_file, "w")
mbligh0a883702010-04-21 01:58:34 +0000712 fh.close()
Prashanth B98509c72014-04-04 16:01:34 -0700713
714
715 def collect_logs(self, remote_src_dir, local_dest_dir, ignore_errors=True):
716 """Copy log directories from a host to a local directory.
717
718 @param remote_src_dir: A destination directory on the host.
719 @param local_dest_dir: A path to a local destination directory.
720 If it doesn't exist it will be created.
721 @param ignore_errors: If True, ignore exceptions.
722
723 @raises OSError: If there were problems creating the local_dest_dir and
724 ignore_errors is False.
725 @raises AutoservRunError, AutotestRunError: If something goes wrong
726 while copying the directories and ignore_errors is False.
727 """
728 locally_created_dest = False
729 if (not os.path.exists(local_dest_dir)
730 or not os.path.isdir(local_dest_dir)):
731 try:
732 os.makedirs(local_dest_dir)
733 locally_created_dest = True
734 except OSError as e:
735 logging.warning('Unable to collect logs from host '
736 '%s: %s', self.hostname, e)
737 if not ignore_errors:
738 raise
739 return
740 try:
741 self.get_file(
742 remote_src_dir, local_dest_dir, preserve_symlinks=True)
743 except (error.AutotestRunError, error.AutoservRunError,
744 error.AutoservSSHTimeout) as e:
745 logging.warning('Collection of %s to local dir %s from host %s '
746 'failed: %s', remote_src_dir, local_dest_dir,
747 self.hostname, e)
748 if locally_created_dest:
749 shutil.rmtree(local_dest_dir, ignore_errors=ignore_errors)
750 if not ignore_errors:
751 raise