blob: 39af9f9a9b100b05dbb4f294a5f6d631238fafcd [file] [log] [blame]
Aviv Keshet07f16242013-10-10 07:54:19 -07001import os, time, socket, shutil, glob, logging, traceback, tempfile
Aviv Keshet53a216a2013-08-27 13:58:46 -07002from autotest_lib.client.common_lib import autotemp, error
jadmanski31c49b72008-10-27 20:44:48 +00003from autotest_lib.server import utils, autotest
mblighe8b93af2009-01-30 00:45:53 +00004from autotest_lib.server.hosts import remote
mblighefccc1b2010-01-11 19:08:42 +00005from autotest_lib.client.common_lib.global_config import global_config
jadmanskica7da372008-10-21 16:26:52 +00006
Aviv Keshet53a216a2013-08-27 13:58:46 -07007# pylint: disable-msg=C0111
jadmanskica7da372008-10-21 16:26:52 +00008
mblighb86bfa12010-02-12 20:22:21 +00009get_value = global_config.get_config_value
10enable_master_ssh = get_value('AUTOSERV', 'enable_master_ssh', type=bool,
11 default=False)
mblighefccc1b2010-01-11 19:08:42 +000012
13
Fang Deng96667ca2013-08-01 17:46:18 -070014class AbstractSSHHost(remote.RemoteHost):
mblighbc9402b2009-12-29 01:15:34 +000015 """
16 This class represents a generic implementation of most of the
jadmanskica7da372008-10-21 16:26:52 +000017 framework necessary for controlling a host via ssh. It implements
18 almost all of the abstract Host methods, except for the core
mblighbc9402b2009-12-29 01:15:34 +000019 Host.run method.
20 """
jadmanskica7da372008-10-21 16:26:52 +000021
jadmanskif6562912008-10-21 17:59:01 +000022 def _initialize(self, hostname, user="root", port=22, password="",
23 *args, **dargs):
24 super(AbstractSSHHost, self)._initialize(hostname=hostname,
25 *args, **dargs)
mbligh6369cf22008-10-24 17:21:57 +000026 self.ip = socket.getaddrinfo(self.hostname, None)[0][4][0]
jadmanskica7da372008-10-21 16:26:52 +000027 self.user = user
28 self.port = port
29 self.password = password
showard6eafb492010-01-15 20:29:06 +000030 self._use_rsync = None
Fang Deng3af66202013-08-16 15:19:25 -070031 self.known_hosts_file = tempfile.mkstemp()[1]
jadmanskica7da372008-10-21 16:26:52 +000032
mblighefccc1b2010-01-11 19:08:42 +000033 """
34 Master SSH connection background job, socket temp directory and socket
35 control path option. If master-SSH is enabled, these fields will be
36 initialized by start_master_ssh when a new SSH connection is initiated.
37 """
38 self.master_ssh_job = None
39 self.master_ssh_tempdir = None
40 self.master_ssh_option = ''
41
showard6eafb492010-01-15 20:29:06 +000042
Fang Deng96667ca2013-08-01 17:46:18 -070043 def make_ssh_command(self, user="root", port=22, opts='',
44 hosts_file='/dev/null',
45 connect_timeout=30, alive_interval=300):
46 base_command = ("/usr/bin/ssh -a -x %s -o StrictHostKeyChecking=no "
47 "-o UserKnownHostsFile=%s -o BatchMode=yes "
48 "-o ConnectTimeout=%d -o ServerAliveInterval=%d "
49 "-l %s -p %d")
50 assert isinstance(connect_timeout, (int, long))
51 assert connect_timeout > 0 # can't disable the timeout
52 return base_command % (opts, hosts_file, connect_timeout,
53 alive_interval, user, port)
54
55
showard6eafb492010-01-15 20:29:06 +000056 def use_rsync(self):
57 if self._use_rsync is not None:
58 return self._use_rsync
59
mblighc9892c02010-01-06 19:02:16 +000060 # Check if rsync is available on the remote host. If it's not,
61 # don't try to use it for any future file transfers.
showard6eafb492010-01-15 20:29:06 +000062 self._use_rsync = self._check_rsync()
63 if not self._use_rsync:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -070064 logging.warning("rsync not available on remote host %s -- disabled",
mblighc9892c02010-01-06 19:02:16 +000065 self.hostname)
Eric Lie0493a42010-11-15 13:05:43 -080066 return self._use_rsync
mblighc9892c02010-01-06 19:02:16 +000067
68
69 def _check_rsync(self):
70 """
71 Check if rsync is available on the remote host.
72 """
73 try:
74 self.run("rsync --version", stdout_tee=None, stderr_tee=None)
75 except error.AutoservRunError:
76 return False
77 return True
78
jadmanskica7da372008-10-21 16:26:52 +000079
showard56176ec2009-10-28 19:52:30 +000080 def _encode_remote_paths(self, paths, escape=True):
mblighbc9402b2009-12-29 01:15:34 +000081 """
82 Given a list of file paths, encodes it as a single remote path, in
83 the style used by rsync and scp.
84 """
showard56176ec2009-10-28 19:52:30 +000085 if escape:
86 paths = [utils.scp_remote_escape(path) for path in paths]
87 return '%s@%s:"%s"' % (self.user, self.hostname, " ".join(paths))
jadmanskica7da372008-10-21 16:26:52 +000088
jadmanskica7da372008-10-21 16:26:52 +000089
mbligh45561782009-05-11 21:14:34 +000090 def _make_rsync_cmd(self, sources, dest, delete_dest, preserve_symlinks):
mblighbc9402b2009-12-29 01:15:34 +000091 """
92 Given a list of source paths and a destination path, produces the
jadmanskid7b79ed2009-01-07 17:19:48 +000093 appropriate rsync command for copying them. Remote paths must be
mblighbc9402b2009-12-29 01:15:34 +000094 pre-encoded.
95 """
Fang Deng96667ca2013-08-01 17:46:18 -070096 ssh_cmd = self.make_ssh_command(user=self.user, port=self.port,
97 opts=self.master_ssh_option,
98 hosts_file=self.known_hosts_file)
jadmanskid7b79ed2009-01-07 17:19:48 +000099 if delete_dest:
100 delete_flag = "--delete"
101 else:
102 delete_flag = ""
mbligh45561782009-05-11 21:14:34 +0000103 if preserve_symlinks:
104 symlink_flag = ""
105 else:
106 symlink_flag = "-L"
Dan Shi06d7fbf2014-02-12 12:34:41 -0800107 command = ("rsync %s %s --timeout=1800 --rsh='%s' -az --no-o --no-g "
David Hendricksb8904182014-06-02 15:22:49 -0700108 "%s \"%s\"")
mbligh45561782009-05-11 21:14:34 +0000109 return command % (symlink_flag, delete_flag, ssh_cmd,
David Hendricksb8904182014-06-02 15:22:49 -0700110 " ".join(['"%s"' % p for p in sources]), dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000111
112
Eric Li861b2d52011-02-04 14:50:35 -0800113 def _make_ssh_cmd(self, cmd):
114 """
115 Create a base ssh command string for the host which can be used
116 to run commands directly on the machine
117 """
Fang Deng96667ca2013-08-01 17:46:18 -0700118 base_cmd = self.make_ssh_command(user=self.user, port=self.port,
119 opts=self.master_ssh_option,
120 hosts_file=self.known_hosts_file)
Eric Li861b2d52011-02-04 14:50:35 -0800121
122 return '%s %s "%s"' % (base_cmd, self.hostname, utils.sh_escape(cmd))
123
jadmanskid7b79ed2009-01-07 17:19:48 +0000124 def _make_scp_cmd(self, sources, dest):
mblighbc9402b2009-12-29 01:15:34 +0000125 """
126 Given a list of source paths and a destination path, produces the
jadmanskid7b79ed2009-01-07 17:19:48 +0000127 appropriate scp command for encoding it. Remote paths must be
mblighbc9402b2009-12-29 01:15:34 +0000128 pre-encoded.
129 """
mblighc0649d62010-01-15 18:15:58 +0000130 command = ("scp -rq %s -o StrictHostKeyChecking=no "
lmraf676f32010-02-04 03:36:26 +0000131 "-o UserKnownHostsFile=%s -P %d %s '%s'")
Fang Deng3af66202013-08-16 15:19:25 -0700132 return command % (self.master_ssh_option, self.known_hosts_file,
mblighefccc1b2010-01-11 19:08:42 +0000133 self.port, " ".join(sources), dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000134
135
136 def _make_rsync_compatible_globs(self, path, is_local):
mblighbc9402b2009-12-29 01:15:34 +0000137 """
138 Given an rsync-style path, returns a list of globbed paths
jadmanskid7b79ed2009-01-07 17:19:48 +0000139 that will hopefully provide equivalent behaviour for scp. Does not
140 support the full range of rsync pattern matching behaviour, only that
141 exposed in the get/send_file interface (trailing slashes).
142
143 The is_local param is flag indicating if the paths should be
mblighbc9402b2009-12-29 01:15:34 +0000144 interpreted as local or remote paths.
145 """
jadmanskid7b79ed2009-01-07 17:19:48 +0000146
147 # non-trailing slash paths should just work
148 if len(path) == 0 or path[-1] != "/":
149 return [path]
150
151 # make a function to test if a pattern matches any files
152 if is_local:
showard56176ec2009-10-28 19:52:30 +0000153 def glob_matches_files(path, pattern):
154 return len(glob.glob(path + pattern)) > 0
jadmanskid7b79ed2009-01-07 17:19:48 +0000155 else:
showard56176ec2009-10-28 19:52:30 +0000156 def glob_matches_files(path, pattern):
157 result = self.run("ls \"%s\"%s" % (utils.sh_escape(path),
158 pattern),
159 stdout_tee=None, ignore_status=True)
jadmanskid7b79ed2009-01-07 17:19:48 +0000160 return result.exit_status == 0
161
162 # take a set of globs that cover all files, and see which are needed
163 patterns = ["*", ".[!.]*"]
showard56176ec2009-10-28 19:52:30 +0000164 patterns = [p for p in patterns if glob_matches_files(path, p)]
jadmanskid7b79ed2009-01-07 17:19:48 +0000165
166 # convert them into a set of paths suitable for the commandline
jadmanskid7b79ed2009-01-07 17:19:48 +0000167 if is_local:
showard56176ec2009-10-28 19:52:30 +0000168 return ["\"%s\"%s" % (utils.sh_escape(path), pattern)
169 for pattern in patterns]
jadmanskid7b79ed2009-01-07 17:19:48 +0000170 else:
showard56176ec2009-10-28 19:52:30 +0000171 return [utils.scp_remote_escape(path) + pattern
172 for pattern in patterns]
jadmanskid7b79ed2009-01-07 17:19:48 +0000173
174
175 def _make_rsync_compatible_source(self, source, is_local):
mblighbc9402b2009-12-29 01:15:34 +0000176 """
177 Applies the same logic as _make_rsync_compatible_globs, but
jadmanskid7b79ed2009-01-07 17:19:48 +0000178 applies it to an entire list of sources, producing a new list of
mblighbc9402b2009-12-29 01:15:34 +0000179 sources, properly quoted.
180 """
jadmanskid7b79ed2009-01-07 17:19:48 +0000181 return sum((self._make_rsync_compatible_globs(path, is_local)
182 for path in source), [])
jadmanskica7da372008-10-21 16:26:52 +0000183
184
mblighfeac0102009-04-28 18:31:12 +0000185 def _set_umask_perms(self, dest):
mblighbc9402b2009-12-29 01:15:34 +0000186 """
187 Given a destination file/dir (recursively) set the permissions on
188 all the files and directories to the max allowed by running umask.
189 """
mblighfeac0102009-04-28 18:31:12 +0000190
191 # now this looks strange but I haven't found a way in Python to _just_
192 # get the umask, apparently the only option is to try to set it
193 umask = os.umask(0)
194 os.umask(umask)
195
196 max_privs = 0777 & ~umask
197
198 def set_file_privs(filename):
Chris Masone567d0d92011-12-19 09:38:30 -0800199 """Sets mode of |filename|. Assumes |filename| exists."""
200 file_stat = os.stat(filename)
mblighfeac0102009-04-28 18:31:12 +0000201
202 file_privs = max_privs
203 # if the original file permissions do not have at least one
204 # executable bit then do not set it anywhere
205 if not file_stat.st_mode & 0111:
206 file_privs &= ~0111
207
208 os.chmod(filename, file_privs)
209
210 # try a bottom-up walk so changes on directory permissions won't cut
211 # our access to the files/directories inside it
212 for root, dirs, files in os.walk(dest, topdown=False):
213 # when setting the privileges we emulate the chmod "X" behaviour
214 # that sets to execute only if it is a directory or any of the
215 # owner/group/other already has execute right
216 for dirname in dirs:
217 os.chmod(os.path.join(root, dirname), max_privs)
218
Chris Masone567d0d92011-12-19 09:38:30 -0800219 # Filter out broken symlinks as we go.
220 for filename in filter(os.path.exists, files):
mblighfeac0102009-04-28 18:31:12 +0000221 set_file_privs(os.path.join(root, filename))
222
223
224 # now set privs for the dest itself
225 if os.path.isdir(dest):
226 os.chmod(dest, max_privs)
227 else:
228 set_file_privs(dest)
229
230
mbligh45561782009-05-11 21:14:34 +0000231 def get_file(self, source, dest, delete_dest=False, preserve_perm=True,
232 preserve_symlinks=False):
jadmanskica7da372008-10-21 16:26:52 +0000233 """
234 Copy files from the remote host to a local path.
235
236 Directories will be copied recursively.
237 If a source component is a directory with a trailing slash,
238 the content of the directory will be copied, otherwise, the
239 directory itself and its content will be copied. This
240 behavior is similar to that of the program 'rsync'.
241
242 Args:
243 source: either
244 1) a single file or directory, as a string
245 2) a list of one or more (possibly mixed)
246 files or directories
247 dest: a file or a directory (if source contains a
248 directory or more than one element, you must
249 supply a directory dest)
mbligh89e258d2008-10-24 13:58:08 +0000250 delete_dest: if this is true, the command will also clear
251 out any old files at dest that are not in the
252 source
mblighfeac0102009-04-28 18:31:12 +0000253 preserve_perm: tells get_file() to try to preserve the sources
254 permissions on files and dirs
mbligh45561782009-05-11 21:14:34 +0000255 preserve_symlinks: try to preserve symlinks instead of
256 transforming them into files/dirs on copy
jadmanskica7da372008-10-21 16:26:52 +0000257
258 Raises:
259 AutoservRunError: the scp command failed
260 """
Simran Basi882f15b2013-10-29 14:59:34 -0700261 logging.debug('get_file. source: %s, dest: %s, delete_dest: %s,'
262 'preserve_perm: %s, preserve_symlinks:%s', source, dest,
263 delete_dest, preserve_perm, preserve_symlinks)
mblighefccc1b2010-01-11 19:08:42 +0000264 # Start a master SSH connection if necessary.
265 self.start_master_ssh()
266
jadmanskica7da372008-10-21 16:26:52 +0000267 if isinstance(source, basestring):
268 source = [source]
jadmanskid7b79ed2009-01-07 17:19:48 +0000269 dest = os.path.abspath(dest)
jadmanskica7da372008-10-21 16:26:52 +0000270
mblighc9892c02010-01-06 19:02:16 +0000271 # If rsync is disabled or fails, try scp.
showard6eafb492010-01-15 20:29:06 +0000272 try_scp = True
273 if self.use_rsync():
Simran Basi882f15b2013-10-29 14:59:34 -0700274 logging.debug('Using Rsync.')
mblighc9892c02010-01-06 19:02:16 +0000275 try:
276 remote_source = self._encode_remote_paths(source)
277 local_dest = utils.sh_escape(dest)
278 rsync = self._make_rsync_cmd([remote_source], local_dest,
279 delete_dest, preserve_symlinks)
280 utils.run(rsync)
showard6eafb492010-01-15 20:29:06 +0000281 try_scp = False
mblighc9892c02010-01-06 19:02:16 +0000282 except error.CmdError, e:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -0700283 logging.warning("trying scp, rsync failed: %s", e)
mblighc9892c02010-01-06 19:02:16 +0000284
285 if try_scp:
Simran Basi882f15b2013-10-29 14:59:34 -0700286 logging.debug('Trying scp.')
jadmanskid7b79ed2009-01-07 17:19:48 +0000287 # scp has no equivalent to --delete, just drop the entire dest dir
288 if delete_dest and os.path.isdir(dest):
289 shutil.rmtree(dest)
290 os.mkdir(dest)
jadmanskica7da372008-10-21 16:26:52 +0000291
jadmanskid7b79ed2009-01-07 17:19:48 +0000292 remote_source = self._make_rsync_compatible_source(source, False)
293 if remote_source:
showard56176ec2009-10-28 19:52:30 +0000294 # _make_rsync_compatible_source() already did the escaping
295 remote_source = self._encode_remote_paths(remote_source,
296 escape=False)
jadmanskid7b79ed2009-01-07 17:19:48 +0000297 local_dest = utils.sh_escape(dest)
jadmanski2583a432009-02-10 23:59:11 +0000298 scp = self._make_scp_cmd([remote_source], local_dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000299 try:
300 utils.run(scp)
301 except error.CmdError, e:
Simran Basi882f15b2013-10-29 14:59:34 -0700302 logging.debug('scp failed: %s', e)
jadmanskid7b79ed2009-01-07 17:19:48 +0000303 raise error.AutoservRunError(e.args[0], e.args[1])
jadmanskica7da372008-10-21 16:26:52 +0000304
mblighfeac0102009-04-28 18:31:12 +0000305 if not preserve_perm:
306 # we have no way to tell scp to not try to preserve the
307 # permissions so set them after copy instead.
308 # for rsync we could use "--no-p --chmod=ugo=rwX" but those
309 # options are only in very recent rsync versions
310 self._set_umask_perms(dest)
311
jadmanskica7da372008-10-21 16:26:52 +0000312
mbligh45561782009-05-11 21:14:34 +0000313 def send_file(self, source, dest, delete_dest=False,
314 preserve_symlinks=False):
jadmanskica7da372008-10-21 16:26:52 +0000315 """
316 Copy files from a local path to the remote host.
317
318 Directories will be copied recursively.
319 If a source component is a directory with a trailing slash,
320 the content of the directory will be copied, otherwise, the
321 directory itself and its content will be copied. This
322 behavior is similar to that of the program 'rsync'.
323
324 Args:
325 source: either
326 1) a single file or directory, as a string
327 2) a list of one or more (possibly mixed)
328 files or directories
329 dest: a file or a directory (if source contains a
330 directory or more than one element, you must
331 supply a directory dest)
mbligh89e258d2008-10-24 13:58:08 +0000332 delete_dest: if this is true, the command will also clear
333 out any old files at dest that are not in the
334 source
mbligh45561782009-05-11 21:14:34 +0000335 preserve_symlinks: controls if symlinks on the source will be
336 copied as such on the destination or transformed into the
337 referenced file/directory
jadmanskica7da372008-10-21 16:26:52 +0000338
339 Raises:
340 AutoservRunError: the scp command failed
341 """
Simran Basi882f15b2013-10-29 14:59:34 -0700342 logging.debug('send_file. source: %s, dest: %s, delete_dest: %s,'
343 'preserve_symlinks:%s', source, dest,
344 delete_dest, preserve_symlinks)
mblighefccc1b2010-01-11 19:08:42 +0000345 # Start a master SSH connection if necessary.
346 self.start_master_ssh()
347
jadmanskica7da372008-10-21 16:26:52 +0000348 if isinstance(source, basestring):
349 source = [source]
jadmanski2583a432009-02-10 23:59:11 +0000350 remote_dest = self._encode_remote_paths([dest])
jadmanskica7da372008-10-21 16:26:52 +0000351
mblighc9892c02010-01-06 19:02:16 +0000352 # If rsync is disabled or fails, try scp.
showard6eafb492010-01-15 20:29:06 +0000353 try_scp = True
354 if self.use_rsync():
Simran Basi882f15b2013-10-29 14:59:34 -0700355 logging.debug('Using Rsync.')
mblighc9892c02010-01-06 19:02:16 +0000356 try:
357 local_sources = [utils.sh_escape(path) for path in source]
358 rsync = self._make_rsync_cmd(local_sources, remote_dest,
359 delete_dest, preserve_symlinks)
360 utils.run(rsync)
showard6eafb492010-01-15 20:29:06 +0000361 try_scp = False
mblighc9892c02010-01-06 19:02:16 +0000362 except error.CmdError, e:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -0700363 logging.warning("trying scp, rsync failed: %s", e)
mblighc9892c02010-01-06 19:02:16 +0000364
365 if try_scp:
Simran Basi882f15b2013-10-29 14:59:34 -0700366 logging.debug('Trying scp.')
jadmanskid7b79ed2009-01-07 17:19:48 +0000367 # scp has no equivalent to --delete, just drop the entire dest dir
368 if delete_dest:
showard27160152009-07-15 14:28:42 +0000369 is_dir = self.run("ls -d %s/" % dest,
jadmanskid7b79ed2009-01-07 17:19:48 +0000370 ignore_status=True).exit_status == 0
371 if is_dir:
372 cmd = "rm -rf %s && mkdir %s"
mbligh5a0ca532009-08-03 16:44:34 +0000373 cmd %= (dest, dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000374 self.run(cmd)
jadmanskica7da372008-10-21 16:26:52 +0000375
jadmanski2583a432009-02-10 23:59:11 +0000376 local_sources = self._make_rsync_compatible_source(source, True)
377 if local_sources:
378 scp = self._make_scp_cmd(local_sources, remote_dest)
jadmanskid7b79ed2009-01-07 17:19:48 +0000379 try:
380 utils.run(scp)
381 except error.CmdError, e:
Simran Basi882f15b2013-10-29 14:59:34 -0700382 logging.debug('scp failed: %s', e)
jadmanskid7b79ed2009-01-07 17:19:48 +0000383 raise error.AutoservRunError(e.args[0], e.args[1])
384
jadmanskica7da372008-10-21 16:26:52 +0000385
beeps46dadc92013-11-07 14:07:10 -0800386 def ssh_ping(self, timeout=60, base_cmd='true'):
beepsadd66d32013-03-04 17:21:51 -0800387 """
388 Pings remote host via ssh.
389
390 @param timeout: Time in seconds before giving up.
391 Defaults to 60 seconds.
beeps46dadc92013-11-07 14:07:10 -0800392 @param base_cmd: The base command to run with the ssh ping.
393 Defaults to true.
beepsadd66d32013-03-04 17:21:51 -0800394 @raise AutoservSSHTimeout: If the ssh ping times out.
395 @raise AutoservSshPermissionDeniedError: If ssh ping fails due to
396 permissions.
397 @raise AutoservSshPingHostError: For other AutoservRunErrors.
398 """
jadmanskica7da372008-10-21 16:26:52 +0000399 try:
beeps46dadc92013-11-07 14:07:10 -0800400 self.run(base_cmd, timeout=timeout, connect_timeout=timeout)
jadmanskica7da372008-10-21 16:26:52 +0000401 except error.AutoservSSHTimeout:
mblighd0e94982009-07-11 00:15:18 +0000402 msg = "Host (ssh) verify timed out (timeout = %d)" % timeout
jadmanskica7da372008-10-21 16:26:52 +0000403 raise error.AutoservSSHTimeout(msg)
mbligh9d738d62009-03-09 21:17:10 +0000404 except error.AutoservSshPermissionDeniedError:
405 #let AutoservSshPermissionDeniedError be visible to the callers
406 raise
jadmanskica7da372008-10-21 16:26:52 +0000407 except error.AutoservRunError, e:
mblighc971c5f2009-06-08 16:48:54 +0000408 # convert the generic AutoservRunError into something more
409 # specific for this context
410 raise error.AutoservSshPingHostError(e.description + '\n' +
411 repr(e.result_obj))
jadmanskica7da372008-10-21 16:26:52 +0000412
413
beeps46dadc92013-11-07 14:07:10 -0800414 def is_up(self, timeout=60, base_cmd='true'):
jadmanskica7da372008-10-21 16:26:52 +0000415 """
beeps46dadc92013-11-07 14:07:10 -0800416 Check if the remote host is up by ssh-ing and running a base command.
jadmanskica7da372008-10-21 16:26:52 +0000417
beepsadd66d32013-03-04 17:21:51 -0800418 @param timeout: timeout in seconds.
beeps46dadc92013-11-07 14:07:10 -0800419 @param base_cmd: a base command to run with ssh. The default is 'true'.
beepsadd66d32013-03-04 17:21:51 -0800420 @returns True if the remote host is up before the timeout expires,
421 False otherwise.
jadmanskica7da372008-10-21 16:26:52 +0000422 """
423 try:
beeps46dadc92013-11-07 14:07:10 -0800424 self.ssh_ping(timeout=timeout, base_cmd=base_cmd)
jadmanskica7da372008-10-21 16:26:52 +0000425 except error.AutoservError:
426 return False
427 else:
428 return True
429
430
431 def wait_up(self, timeout=None):
432 """
433 Wait until the remote host is up or the timeout expires.
434
435 In fact, it will wait until an ssh connection to the remote
436 host can be established, and getty is running.
437
jadmanskic0354912010-01-12 15:57:29 +0000438 @param timeout time limit in seconds before returning even
439 if the host is not up.
jadmanskica7da372008-10-21 16:26:52 +0000440
beepsadd66d32013-03-04 17:21:51 -0800441 @returns True if the host was found to be up before the timeout expires,
442 False otherwise
jadmanskica7da372008-10-21 16:26:52 +0000443 """
444 if timeout:
beeps46dadc92013-11-07 14:07:10 -0800445 current_time = int(time.time())
446 end_time = current_time + timeout
jadmanskica7da372008-10-21 16:26:52 +0000447
beepsadd66d32013-03-04 17:21:51 -0800448 while not timeout or current_time < end_time:
449 if self.is_up(timeout=end_time - current_time):
jadmanskica7da372008-10-21 16:26:52 +0000450 try:
451 if self.are_wait_up_processes_up():
jadmanski7ebac3d2010-06-17 16:06:31 +0000452 logging.debug('Host %s is now up', self.hostname)
jadmanskica7da372008-10-21 16:26:52 +0000453 return True
454 except error.AutoservError:
455 pass
456 time.sleep(1)
beeps46dadc92013-11-07 14:07:10 -0800457 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000458
jadmanski7ebac3d2010-06-17 16:06:31 +0000459 logging.debug('Host %s is still down after waiting %d seconds',
460 self.hostname, int(timeout + time.time() - end_time))
jadmanskica7da372008-10-21 16:26:52 +0000461 return False
462
463
jadmanskic0354912010-01-12 15:57:29 +0000464 def wait_down(self, timeout=None, warning_timer=None, old_boot_id=None):
jadmanskica7da372008-10-21 16:26:52 +0000465 """
466 Wait until the remote host is down or the timeout expires.
467
jadmanskic0354912010-01-12 15:57:29 +0000468 If old_boot_id is provided, this will wait until either the machine
469 is unpingable or self.get_boot_id() returns a value different from
470 old_boot_id. If the boot_id value has changed then the function
471 returns true under the assumption that the machine has shut down
472 and has now already come back up.
jadmanskica7da372008-10-21 16:26:52 +0000473
jadmanskic0354912010-01-12 15:57:29 +0000474 If old_boot_id is None then until the machine becomes unreachable the
475 method assumes the machine has not yet shut down.
jadmanskica7da372008-10-21 16:26:52 +0000476
beepsadd66d32013-03-04 17:21:51 -0800477 Based on this definition, the 4 possible permutations of timeout
478 and old_boot_id are:
479 1. timeout and old_boot_id: wait timeout seconds for either the
480 host to become unpingable, or the boot id
481 to change. In the latter case we've rebooted
482 and in the former case we've only shutdown,
483 but both cases return True.
484 2. only timeout: wait timeout seconds for the host to become unpingable.
485 If the host remains pingable throughout timeout seconds
486 we return False.
487 3. only old_boot_id: wait forever until either the host becomes
488 unpingable or the boot_id changes. Return true
489 when either of those conditions are met.
490 4. not timeout, not old_boot_id: wait forever till the host becomes
491 unpingable.
492
jadmanskic0354912010-01-12 15:57:29 +0000493 @param timeout Time limit in seconds before returning even
494 if the host is still up.
495 @param warning_timer Time limit in seconds that will generate
496 a warning if the host is not down yet.
497 @param old_boot_id A string containing the result of self.get_boot_id()
498 prior to the host being told to shut down. Can be None if this is
499 not available.
500
501 @returns True if the host was found to be down, False otherwise
jadmanskica7da372008-10-21 16:26:52 +0000502 """
mblighe5e3cf22010-05-27 23:33:14 +0000503 #TODO: there is currently no way to distinguish between knowing
504 #TODO: boot_id was unsupported and not knowing the boot_id.
beeps46dadc92013-11-07 14:07:10 -0800505 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000506 if timeout:
mbligh2ed998f2009-04-08 21:03:47 +0000507 end_time = current_time + timeout
jadmanskica7da372008-10-21 16:26:52 +0000508
mbligh2ed998f2009-04-08 21:03:47 +0000509 if warning_timer:
510 warn_time = current_time + warning_timer
511
jadmanskic0354912010-01-12 15:57:29 +0000512 if old_boot_id is not None:
513 logging.debug('Host %s pre-shutdown boot_id is %s',
514 self.hostname, old_boot_id)
515
beepsadd66d32013-03-04 17:21:51 -0800516 # Impose semi real-time deadline constraints, since some clients
517 # (eg: watchdog timer tests) expect strict checking of time elapsed.
518 # Each iteration of this loop is treated as though it atomically
519 # completes within current_time, this is needed because if we used
520 # inline time.time() calls instead then the following could happen:
521 #
522 # while not timeout or time.time() < end_time: [23 < 30]
523 # some code. [takes 10 secs]
524 # try:
525 # new_boot_id = self.get_boot_id(timeout=end_time - time.time())
526 # [30 - 33]
527 # The last step will lead to a return True, when in fact the machine
528 # went down at 32 seconds (>30). Hence we need to pass get_boot_id
529 # the same time that allowed us into that iteration of the loop.
mbligh2ed998f2009-04-08 21:03:47 +0000530 while not timeout or current_time < end_time:
jadmanskic0354912010-01-12 15:57:29 +0000531 try:
beeps46dadc92013-11-07 14:07:10 -0800532 new_boot_id = self.get_boot_id(timeout=end_time-current_time)
mblighdbc7e4a2010-01-15 20:34:20 +0000533 except error.AutoservError:
jadmanskic0354912010-01-12 15:57:29 +0000534 logging.debug('Host %s is now unreachable over ssh, is down',
535 self.hostname)
jadmanskica7da372008-10-21 16:26:52 +0000536 return True
jadmanskic0354912010-01-12 15:57:29 +0000537 else:
538 # if the machine is up but the boot_id value has changed from
539 # old boot id, then we can assume the machine has gone down
540 # and then already come back up
541 if old_boot_id is not None and old_boot_id != new_boot_id:
542 logging.debug('Host %s now has boot_id %s and so must '
543 'have rebooted', self.hostname, new_boot_id)
544 return True
mbligh2ed998f2009-04-08 21:03:47 +0000545
546 if warning_timer and current_time > warn_time:
Scott Zawalskic86fdeb2013-10-23 10:24:04 -0400547 self.record("INFO", None, "shutdown",
mbligh2ed998f2009-04-08 21:03:47 +0000548 "Shutdown took longer than %ds" % warning_timer)
549 # Print the warning only once.
550 warning_timer = None
mbligha4464402009-04-17 20:13:41 +0000551 # If a machine is stuck switching runlevels
552 # This may cause the machine to reboot.
553 self.run('kill -HUP 1', ignore_status=True)
mbligh2ed998f2009-04-08 21:03:47 +0000554
jadmanskica7da372008-10-21 16:26:52 +0000555 time.sleep(1)
beeps46dadc92013-11-07 14:07:10 -0800556 current_time = int(time.time())
jadmanskica7da372008-10-21 16:26:52 +0000557
558 return False
jadmanskif6562912008-10-21 17:59:01 +0000559
mbligha0a27592009-01-24 01:41:36 +0000560
jadmanskif6562912008-10-21 17:59:01 +0000561 # tunable constants for the verify & repair code
mblighb86bfa12010-02-12 20:22:21 +0000562 AUTOTEST_GB_DISKSPACE_REQUIRED = get_value("SERVER",
563 "gb_diskspace_required",
Fang Deng6b05f5b2013-03-20 13:42:11 -0700564 type=float,
565 default=20.0)
mbligha0a27592009-01-24 01:41:36 +0000566
jadmanskif6562912008-10-21 17:59:01 +0000567
showardca572982009-09-18 21:20:01 +0000568 def verify_connectivity(self):
569 super(AbstractSSHHost, self).verify_connectivity()
jadmanskif6562912008-10-21 17:59:01 +0000570
showardb18134f2009-03-20 20:52:18 +0000571 logging.info('Pinging host ' + self.hostname)
jadmanskif6562912008-10-21 17:59:01 +0000572 self.ssh_ping()
mbligh2ba7ab02009-08-24 22:09:26 +0000573 logging.info("Host (ssh) %s is alive", self.hostname)
jadmanskif6562912008-10-21 17:59:01 +0000574
jadmanski80deb752009-01-21 17:14:16 +0000575 if self.is_shutting_down():
mblighc971c5f2009-06-08 16:48:54 +0000576 raise error.AutoservHostIsShuttingDownError("Host is shutting down")
jadmanski80deb752009-01-21 17:14:16 +0000577
mblighb49b5232009-02-12 21:54:49 +0000578
showardca572982009-09-18 21:20:01 +0000579 def verify_software(self):
580 super(AbstractSSHHost, self).verify_software()
jadmanskif6562912008-10-21 17:59:01 +0000581 try:
showardad812bf2009-10-20 23:49:56 +0000582 self.check_diskspace(autotest.Autotest.get_install_dir(self),
583 self.AUTOTEST_GB_DISKSPACE_REQUIRED)
jadmanskif6562912008-10-21 17:59:01 +0000584 except error.AutoservHostError:
585 raise # only want to raise if it's a space issue
showardad812bf2009-10-20 23:49:56 +0000586 except autotest.AutodirNotFoundError:
showardca572982009-09-18 21:20:01 +0000587 # autotest dir may not exist, etc. ignore
588 logging.debug('autodir space check exception, this is probably '
589 'safe to ignore\n' + traceback.format_exc())
mblighefccc1b2010-01-11 19:08:42 +0000590
591
592 def close(self):
593 super(AbstractSSHHost, self).close()
594 self._cleanup_master_ssh()
Fang Deng3af66202013-08-16 15:19:25 -0700595 os.remove(self.known_hosts_file)
mblighefccc1b2010-01-11 19:08:42 +0000596
597
598 def _cleanup_master_ssh(self):
599 """
600 Release all resources (process, temporary directory) used by an active
601 master SSH connection.
602 """
603 # If a master SSH connection is running, kill it.
604 if self.master_ssh_job is not None:
Aviv Keshet46250752013-08-27 15:52:06 -0700605 logging.debug('Nuking master_ssh_job.')
mblighefccc1b2010-01-11 19:08:42 +0000606 utils.nuke_subprocess(self.master_ssh_job.sp)
607 self.master_ssh_job = None
608
609 # Remove the temporary directory for the master SSH socket.
610 if self.master_ssh_tempdir is not None:
Aviv Keshet46250752013-08-27 15:52:06 -0700611 logging.debug('Cleaning master_ssh_tempdir.')
mblighefccc1b2010-01-11 19:08:42 +0000612 self.master_ssh_tempdir.clean()
613 self.master_ssh_tempdir = None
614 self.master_ssh_option = ''
615
616
Aviv Keshet0749a822013-10-17 09:53:26 -0700617 def start_master_ssh(self, timeout=5):
mblighefccc1b2010-01-11 19:08:42 +0000618 """
619 Called whenever a slave SSH connection needs to be initiated (e.g., by
620 run, rsync, scp). If master SSH support is enabled and a master SSH
621 connection is not active already, start a new one in the background.
622 Also, cleanup any zombie master SSH connections (e.g., dead due to
623 reboot).
Aviv Keshet0749a822013-10-17 09:53:26 -0700624
625 timeout: timeout in seconds (default 5) to wait for master ssh
626 connection to be established. If timeout is reached, a
627 warning message is logged, but no other action is taken.
mblighefccc1b2010-01-11 19:08:42 +0000628 """
629 if not enable_master_ssh:
630 return
631
632 # If a previously started master SSH connection is not running
633 # anymore, it needs to be cleaned up and then restarted.
634 if self.master_ssh_job is not None:
Aviv Keshet49b4ed02013-09-05 14:48:21 -0700635 socket_path = os.path.join(self.master_ssh_tempdir.name, 'socket')
636 if (not os.path.exists(socket_path) or
637 self.master_ssh_job.sp.poll() is not None):
Aviv Keshet982fde12013-09-06 11:24:10 -0700638 if self.master_ssh_job.sp.poll() is None:
639 logging.warning('Master ssh connection socket file '
640 'was missing while its subprocess was '
Aviv Keshet3ae1a5f2013-10-09 12:22:56 -0400641 'still running.')
642 if os.path.exists(self.master_ssh_tempdir.name):
643 logging.warning('However, the socket file temporary '
644 'directory still exists.')
Aviv Keshet07f16242013-10-10 07:54:19 -0700645
Aviv Keshet3ae1a5f2013-10-09 12:22:56 -0400646 logging.warning('Info on defunct master ssh ps below.')
Aviv Keshet44c609e2013-09-06 12:57:49 -0700647 master_pid = str(self.master_ssh_job.sp.pid)
Aviv Keshet07f16242013-10-10 07:54:19 -0700648 ps_output = utils.run(['ps', '-Fww', master_pid],
649 ignore_status=True).stdout
650 logging.warning('Master ssh connection ps info: %s',
Aviv Keshet44c609e2013-09-06 12:57:49 -0700651 ps_output)
Aviv Keshet07f16242013-10-10 07:54:19 -0700652 lsof_output = utils.run(['lsof', '-p', master_pid],
653 ignore_status=True).stdout
654 logging.warning('Master ssh connection lsof info: %s',
655 lsof_output)
Aviv Keshet44c609e2013-09-06 12:57:49 -0700656
mblighefccc1b2010-01-11 19:08:42 +0000657 logging.info("Master ssh connection to %s is down.",
658 self.hostname)
659 self._cleanup_master_ssh()
660
661 # Start a new master SSH connection.
662 if self.master_ssh_job is None:
663 # Create a shared socket in a temp location.
664 self.master_ssh_tempdir = autotemp.tempdir(unique_id='ssh-master')
665 self.master_ssh_option = ("-o ControlPath=%s/socket" %
666 self.master_ssh_tempdir.name)
667
668 # Start the master SSH connection in the background.
mbligh5644c122010-01-29 17:43:26 +0000669 master_cmd = self.ssh_command(options="-N -o ControlMaster=yes")
Aviv Keshet31b9ec82013-10-15 13:58:13 -0400670 logging.debug("System load: %s", utils.run(['uptime']).stdout)
Aviv Keshet46250752013-08-27 15:52:06 -0700671 logging.info("Starting master ssh connection '%s'", master_cmd)
Aviv Keshet53a216a2013-08-27 13:58:46 -0700672 self.master_ssh_job = utils.BgJob(master_cmd,
Aviv Keshet46250752013-08-27 15:52:06 -0700673 nickname='master-ssh',
674 no_pipes=True)
Aviv Keshet0749a822013-10-17 09:53:26 -0700675 # To prevent a race between the the master ssh connection startup
676 # and its first attempted use, wait for socket file to exist before
677 # returning.
678 end_time = time.time() + timeout
679 socket_file_path = os.path.join(self.master_ssh_tempdir.name,
680 'socket')
681 while time.time() < end_time:
682 if os.path.exists(socket_file_path):
683 break
684 time.sleep(.2)
685 else:
Ilja H. Friedel04be2bd2014-05-07 21:29:59 -0700686 logging.warning('Timed out waiting for master-ssh connection '
Aviv Keshet0749a822013-10-17 09:53:26 -0700687 'to be established.')
mbligh0a883702010-04-21 01:58:34 +0000688
689
690 def clear_known_hosts(self):
691 """Clears out the temporary ssh known_hosts file.
692
693 This is useful if the test SSHes to the machine, then reinstalls it,
694 then SSHes to it again. It can be called after the reinstall to
695 reduce the spam in the logs.
696 """
697 logging.info("Clearing known hosts for host '%s', file '%s'.",
Fang Deng3af66202013-08-16 15:19:25 -0700698 self.hostname, self.known_hosts_file)
mbligh0a883702010-04-21 01:58:34 +0000699 # Clear out the file by opening it for writing and then closing.
Fang Deng3af66202013-08-16 15:19:25 -0700700 fh = open(self.known_hosts_file, "w")
mbligh0a883702010-04-21 01:58:34 +0000701 fh.close()
Prashanth B98509c72014-04-04 16:01:34 -0700702
703
704 def collect_logs(self, remote_src_dir, local_dest_dir, ignore_errors=True):
705 """Copy log directories from a host to a local directory.
706
707 @param remote_src_dir: A destination directory on the host.
708 @param local_dest_dir: A path to a local destination directory.
709 If it doesn't exist it will be created.
710 @param ignore_errors: If True, ignore exceptions.
711
712 @raises OSError: If there were problems creating the local_dest_dir and
713 ignore_errors is False.
714 @raises AutoservRunError, AutotestRunError: If something goes wrong
715 while copying the directories and ignore_errors is False.
716 """
717 locally_created_dest = False
718 if (not os.path.exists(local_dest_dir)
719 or not os.path.isdir(local_dest_dir)):
720 try:
721 os.makedirs(local_dest_dir)
722 locally_created_dest = True
723 except OSError as e:
724 logging.warning('Unable to collect logs from host '
725 '%s: %s', self.hostname, e)
726 if not ignore_errors:
727 raise
728 return
729 try:
730 self.get_file(
731 remote_src_dir, local_dest_dir, preserve_symlinks=True)
732 except (error.AutotestRunError, error.AutoservRunError,
733 error.AutoservSSHTimeout) as e:
734 logging.warning('Collection of %s to local dir %s from host %s '
735 'failed: %s', remote_src_dir, local_dest_dir,
736 self.hostname, e)
737 if locally_created_dest:
738 shutil.rmtree(local_dest_dir, ignore_errors=ignore_errors)
739 if not ignore_errors:
740 raise