X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=src%2Fnepi%2Fresources%2Flinux%2Fnode.py;h=b0d63c93f5a3e6ab23cdece8f0dc3d2dafa13435;hb=336762742278c5c0df8157a90ef0126ca8b83a42;hp=214c4ef05b05ea6a674acd2f60784d4419e4787d;hpb=0cff9603cef884887e19b6f594745284e54a2b19;p=nepi.git diff --git a/src/nepi/resources/linux/node.py b/src/nepi/resources/linux/node.py index 214c4ef0..b0d63c93 100644 --- a/src/nepi/resources/linux/node.py +++ b/src/nepi/resources/linux/node.py @@ -3,9 +3,8 @@ # Copyright (C) 2013 INRIA # # This program is free software: you can redistribute it and/or modify -# it under the terms of the GNU General Public License as published by -# the Free Software Foundation, either version 3 of the License, or -# (at your option) any later version. +# it under the terms of the GNU General Public License version 2 as +# published by the Free Software Foundation; # # This program is distributed in the hope that it will be useful, # but WITHOUT ANY WARRANTY; without even the implied warranty of @@ -19,7 +18,7 @@ from nepi.execution.attribute import Attribute, Flags, Types from nepi.execution.resource import ResourceManager, clsinit_copy, \ - ResourceState, reschedule_delay + ResourceState from nepi.resources.linux import rpmfuncs, debfuncs from nepi.util import sshfuncs, execfuncs from nepi.util.sshfuncs import ProcStatus @@ -50,12 +49,12 @@ class OSType: """ Supported flavors of Linux OS """ - FEDORA_8 = "f8" - FEDORA_12 = "f12" - FEDORA_14 = "f14" - FEDORA = "fedora" - UBUNTU = "ubuntu" - DEBIAN = "debian" + DEBIAN = 1 + UBUNTU = 1 << 1 + FEDORA = 1 << 2 + FEDORA_8 = 1 << 3 | FEDORA + FEDORA_12 = 1 << 4 | FEDORA + FEDORA_14 = 1 << 5 | FEDORA @clsinit_copy class LinuxNode(ResourceManager): @@ -141,10 +140,10 @@ class LinuxNode(ResourceManager): source compilation, file download, etc) """ - _rtype = "LinuxNode" + _rtype = "linux::Node" _help = "Controls Linux host machines ( either localhost or a host " \ "that can be accessed using a SSH key)" - _backend_type = "linux" + _platform = "linux" @classmethod def _register_attributes(cls): @@ -184,6 +183,13 @@ class LinuxNode(ResourceManager): default = False, flags = Flags.Design) + clean_processes_after = Attribute("cleanProcessesAfter", + """Kill all running processes after starting experiment +This might be dangerous when using user root""", + type = Types.Bool, + default = True, + flags = Flags.Design) + tear_down = Attribute("tearDown", "Bash script to be executed before " + \ "releasing the resource", flags = Flags.Design) @@ -194,6 +200,10 @@ class LinuxNode(ResourceManager): gateway = Attribute("gateway", "Hostname of the gateway machine", flags = Flags.Design) + ip = Attribute("ip", "Linux host public IP address. " + "Must not be modified by the user unless hostname is 'localhost'", + flags = Flags.Design) + cls._register_attribute(hostname) cls._register_attribute(username) cls._register_attribute(port) @@ -203,9 +213,11 @@ class LinuxNode(ResourceManager): cls._register_attribute(clean_home) cls._register_attribute(clean_experiment) cls._register_attribute(clean_processes) + cls._register_attribute(clean_processes_after) cls._register_attribute(tear_down) cls._register_attribute(gateway_user) cls._register_attribute(gateway) + cls._register_attribute(ip) def __init__(self, ec, guid): super(LinuxNode, self).__init__(ec, guid) @@ -233,9 +245,13 @@ class LinuxNode(ResourceManager): home = os.path.join(self._home_dir, home) return home + @property + def nepi_home(self): + return os.path.join(self.home_dir, ".nepi") + @property def usr_dir(self): - return os.path.join(self.home_dir, "nepi-usr") + return os.path.join(self.nepi_home, "nepi-usr") @property def lib_dir(self): @@ -255,7 +271,7 @@ class LinuxNode(ResourceManager): @property def exp_dir(self): - return os.path.join(self.home_dir, "nepi-exp") + return os.path.join(self.nepi_home, "nepi-exp") @property def exp_home(self): @@ -274,26 +290,25 @@ class LinuxNode(ResourceManager): if self._os: return self._os - if self.get("hostname") not in ["localhost", "127.0.0.1"] and \ - not self.get("username"): + if not self.localhost and not self.get("username"): msg = "Can't resolve OS, insufficient data " self.error(msg) raise RuntimeError, msg out = self.get_os() - if out.find("Fedora release 8") == 0: - self._os = OSType.FEDORA_8 - elif out.find("Fedora release 12") == 0: - self._os = OSType.FEDORA_12 - elif out.find("Fedora release 14") == 0: - self._os = OSType.FEDORA_14 - elif out.find("Fedora release") == 0: - self._os = OSType.FEDORA - elif out.find("Debian") == 0: + if out.find("Debian") == 0: self._os = OSType.DEBIAN elif out.find("Ubuntu") ==0: self._os = OSType.UBUNTU + elif out.find("Fedora release") == 0: + self._os = OSType.FEDORA + if out.find("Fedora release 8") == 0: + self._os = OSType.FEDORA_8 + elif out.find("Fedora release 12") == 0: + self._os = OSType.FEDORA_12 + elif out.find("Fedora release 14") == 0: + self._os = OSType.FEDORA_14 else: msg = "Unsupported OS" self.error(msg, out) @@ -320,16 +335,15 @@ class LinuxNode(ResourceManager): @property def use_deb(self): - return self.os in [OSType.DEBIAN, OSType.UBUNTU] + return (self.os & (OSType.DEBIAN|OSType.UBUNTU)) @property def use_rpm(self): - return self.os in [OSType.FEDORA_12, OSType.FEDORA_14, OSType.FEDORA_8, - OSType.FEDORA] + return (self.os & OSType.FEDORA) @property def localhost(self): - return self.get("hostname") in ['localhost', '127.0.0.7', '::1'] + return self.get("hostname") in ['localhost', '127.0.0.1', '::1'] def do_provision(self): # check if host is alive @@ -349,14 +363,24 @@ class LinuxNode(ResourceManager): if self.get("cleanExperiment"): self.clean_experiment() - # Create shared directory structure - self.mkdir(self.lib_dir) - self.mkdir(self.bin_dir) - self.mkdir(self.src_dir) - self.mkdir(self.share_dir) + # Create shared directory structure and node home directory + paths = [self.lib_dir, + self.bin_dir, + self.src_dir, + self.share_dir, + self.node_home] - # Create experiment node home directory - self.mkdir(self.node_home) + self.mkdir(paths) + + # Get Public IP address if possible + if not self.get("ip"): + try: + ip = sshfuncs.gethostbyname(self.get("hostname")) + self.set("ip", ip) + except: + if self.get("gateway") is None: + msg = "Local DNS can not resolve hostname %s" % self.get("hostname") + self.error(msg) super(LinuxNode, self).do_provision() @@ -372,7 +396,7 @@ class LinuxNode(ResourceManager): ifaces = self.get_connected(LinuxInterface.get_rtype()) for iface in ifaces: if iface.state < ResourceState.READY: - self.ec.schedule(reschedule_delay, self.deploy) + self.ec.schedule(self.reschedule_delay, self.deploy) return super(LinuxNode, self).do_deploy() @@ -383,14 +407,15 @@ class LinuxNode(ResourceManager): # Node needs to wait until all associated RMs are released # before it can be released if rm.state != ResourceState.RELEASED: - self.ec.schedule(reschedule_delay, self.release) + self.ec.schedule(self.reschedule_delay, self.release) return tear_down = self.get("tearDown") if tear_down: self.execute(tear_down) - self.clean_processes() + if self.get("cleanProcessesAfter"): + self.clean_processes() super(LinuxNode, self).do_release() @@ -400,13 +425,13 @@ class LinuxNode(ResourceManager): def clean_processes(self): self.info("Cleaning up processes") - - if self.get("hostname") in ["localhost", "127.0.0.2"]: + + if self.localhost: return if self.get("username") != 'root': cmd = ("sudo -S killall tcpdump || /bin/true ; " + - "sudo -S kill $(ps aux | grep '[n]epi' | awk '{print $2}') || /bin/true ; " + + "sudo -S kill -9 $(ps aux | grep '[.]nepi' | awk '{print $2}') || /bin/true ; " + "sudo -S killall -u %s || /bin/true ; " % self.get("username")) else: if self.state >= ResourceState.READY: @@ -415,18 +440,22 @@ class LinuxNode(ResourceManager): pids_temp = dict() ps_aux = "ps aux |awk '{print $2,$11}'" (out, err), proc = self.execute(ps_aux) - for line in out.strip().split("\n"): - parts = line.strip().split(" ") - pids_temp[parts[0]] = parts[1] - kill_pids = set(pids_temp.items()) - set(pids.items()) - kill_pids = ' '.join(dict(kill_pids).keys()) - - cmd = ("killall tcpdump || /bin/true ; " + - "kill $(ps aux | grep '[n]epi' | awk '{print $2}') || /bin/true ; " + - "kill %s || /bin/true ; " % kill_pids) + if len(out) != 0: + for line in out.strip().split("\n"): + parts = line.strip().split(" ") + pids_temp[parts[0]] = parts[1] + kill_pids = set(pids_temp.items()) - set(pids.items()) + kill_pids = ' '.join(dict(kill_pids).keys()) + + cmd = ("killall tcpdump || /bin/true ; " + + "kill $(ps aux | grep '[.]nepi' | awk '{print $2}') || /bin/true ; " + + "kill %s || /bin/true ; " % kill_pids) + else: + cmd = ("killall tcpdump || /bin/true ; " + + "kill $(ps aux | grep '[.]nepi' | awk '{print $2}') || /bin/true ; ") else: cmd = ("killall tcpdump || /bin/true ; " + - "kill $(ps aux | grep '[n]epi' | awk '{print $2}') || /bin/true ; ") + "kill $(ps aux | grep '[.]nepi' | awk '{print $2}') || /bin/true ; ") (out, err), proc = self.execute(cmd, retry = 1, with_lock = True) @@ -435,7 +464,7 @@ class LinuxNode(ResourceManager): """ self.info("Cleaning up home") - cmd = "cd %s ; find . -maxdepth 1 \( -name 'nepi-usr' -o -name 'nepi-exp' \) -execdir rm -rf {} + " % ( + cmd = "cd %s ; find . -maxdepth 1 -name \.nepi -execdir rm -rf {} + " % ( self.home_dir ) return self.execute(cmd, with_lock = True) @@ -532,7 +561,8 @@ class LinuxNode(ResourceManager): stdout = 'stdout', stderr = 'stderr', sudo = False, - tty = False): + tty = False, + strict_host_checking = False): self.debug("Running command '%s'" % command) @@ -563,7 +593,8 @@ class LinuxNode(ResourceManager): agent = True, identity = self.get("identity"), server_key = self.get("serverKey"), - tty = tty + tty = tty, + strict_host_checking = strict_host_checking ) return (out, err), proc @@ -582,7 +613,8 @@ class LinuxNode(ResourceManager): gw = self.get("gateway"), agent = True, identity = self.get("identity"), - server_key = self.get("serverKey") + server_key = self.get("serverKey"), + strict_host_checking = False ) return pidtuple @@ -601,7 +633,8 @@ class LinuxNode(ResourceManager): gw = self.get("gateway"), agent = True, identity = self.get("identity"), - server_key = self.get("serverKey") + server_key = self.get("serverKey"), + strict_host_checking = False ) return status @@ -626,7 +659,8 @@ class LinuxNode(ResourceManager): agent = True, sudo = sudo, identity = self.get("identity"), - server_key = self.get("serverKey") + server_key = self.get("serverKey"), + strict_host_checking = False ) return (out, err), proc @@ -698,7 +732,8 @@ class LinuxNode(ResourceManager): if err: msg = " Failed to upload files - src: %s dst: %s" % (";".join(src), dst) self.error(msg, out, err) - + + msg = "%s out: %s err: %s" % (msg, out, err) if raise_on_error: raise RuntimeError, msg @@ -784,27 +819,45 @@ class LinuxNode(ResourceManager): return (out, err), proc - def mkdir(self, path, clean = False): + def mkdir(self, paths, clean = False): + """ Paths is either a single remote directory path to create, + or a list of directories to create. + """ if clean: - self.rmdir(path) + self.rmdir(paths) + + if isinstance(paths, str): + paths = [paths] + + cmd = " ; ".join(map(lambda path: "mkdir -p %s" % path, paths)) + + return self.execute(cmd, with_lock = True) + + def rmdir(self, paths): + """ Paths is either a single remote directory path to delete, + or a list of directories to delete. + """ - return self.execute("mkdir -p %s" % path, with_lock = True) + if isinstance(paths, str): + paths = [paths] - def rmdir(self, path): - return self.execute("rm -rf %s" % path, with_lock = True) + cmd = " ; ".join(map(lambda path: "rm -rf %s" % path, paths)) + + return self.execute(cmd, with_lock = True) def run_and_wait(self, command, home, - shfile = "cmd.sh", - env = None, - overwrite = True, - pidfile = "pidfile", - ecodefile = "exitcode", - stdin = None, - stdout = "stdout", - stderr = "stderr", - sudo = False, - tty = False, - raise_on_error = True): + shfile="cmd.sh", + env=None, + overwrite=True, + wait_run=True, + pidfile="pidfile", + ecodefile="exitcode", + stdin=None, + stdout="stdout", + stderr="stderr", + sudo=False, + tty=False, + raise_on_error=True): """ Uploads the 'command' to a bash script in the host. Then runs the script detached in background in the host, and @@ -843,25 +896,26 @@ class LinuxNode(ResourceManager): pidfile = pidfile, raise_on_error = raise_on_error) - # wait until command finishes to execute - self.wait_run(pid, ppid) - - (eout, err), proc = self.check_errors(home, - ecodefile = ecodefile, - stderr = stderr) + if wait_run: + # wait until command finishes to execute + self.wait_run(pid, ppid) + + (eout, err), proc = self.check_errors(home, + ecodefile = ecodefile, + stderr = stderr) - # Out is what was written in the stderr file - if err: - msg = " Failed to run command '%s' " % command - self.error(msg, eout, err) + # Out is what was written in the stderr file + if err: + msg = " Failed to run command '%s' " % command + self.error(msg, eout, err) - if raise_on_error: - raise RuntimeError, msg + if raise_on_error: + raise RuntimeError, msg (out, oerr), proc = self.check_output(home, stdout) return (out, err), proc - + def exitcode(self, home, ecodefile = "exitcode"): """ Get the exit code of an application. @@ -885,10 +939,10 @@ class LinuxNode(ResourceManager): return ExitCode.ERROR def upload_command(self, command, - shfile = "cmd.sh", - ecodefile = "exitcode", - overwrite = True, - env = None): + shfile="cmd.sh", + ecodefile="exitcode", + overwrite=True, + env=None): """ Saves the command as a bash script file in the remote host, and forces to save the exit code of the command execution to the ecodefile """ @@ -908,9 +962,9 @@ class LinuxNode(ResourceManager): # Add environ to command command = environ + command - return self.upload(command, shfile, text = True, overwrite = overwrite) + return self.upload(command, shfile, text=True, overwrite=overwrite) - def format_environment(self, env, inline = False): + def format_environment(self, env, inline=False): """ Formats the environment variables for a command to be executed either as an inline command (i.e. export PYTHONPATH=src/..; export LALAL= ..;python script.py) or @@ -1058,8 +1112,7 @@ class LinuxNode(ResourceManager): """ Removes files that already exist in the Linux host from src list """ # construct a dictionary with { dst: src } - dests = dict(map( - lambda s: (os.path.join(dst, os.path.basename(s)), s ), s)) \ + dests = dict(map(lambda s: (os.path.join(dst, os.path.basename(s)), s), src)) \ if len(src) > 1 else dict({dst: src[0]}) command = []