Extend the server-side profiler code to handle reboots as well. Sadly,
this is a bit of a kludge, but I can't see any better solution without
redesigning the client-side profiler interface first.
I handle a reboot by pulling back any existing profiler data right
after a reboot and then re-starting the profiler. This generally works
because most of the profilers are idempotent; either they write into
different files in each run, or they just continuously append logs
to a log file of some kind and so concatenating all their output
is good enough.
In general there's unfortunately no reliable way to signal the
profilers before a reboot occurs, we can only deal with it after the
fact, so we don't have the option of adding some sort of
save-and-restore methods to the client-side interface that could be
used to deal with this.
Right now we simply don't have any real use cases anyway where we want
to run a profiler on top of a reboot test where the profiler requires
special handling after a reboot. Until we actually have an example of
that to work with, trying to implement support for that situation
may just be overdesigning things anyway.
Risk: Medium
Visibility: Server-side profilers will be restarted after a reboot
during a test.
Signed-off-by: John Admanski <jadmanski@google.com>
git-svn-id: http://test.kernel.org/svn/autotest/trunk@2526 592f7852-d20e-0410-864c-8624ca9c26a4
diff --git a/server/profiler.py b/server/profiler.py
index 63b2e4d..df3deac 100644
--- a/server/profiler.py
+++ b/server/profiler.py
@@ -1,4 +1,5 @@
-import os, itertools
+import os, itertools, shutil, tempfile
+from autotest_lib.client.common_lib import utils, error
from autotest_lib.server import autotest
@@ -61,6 +62,7 @@
self.job = job
self.name = profiler_name
self.installed_hosts = {}
+ self.current_test = None
def _install(self):
@@ -91,39 +93,88 @@
# the actual setup happens lazily at start()
- def _signal_clients(self, command):
- """ Signal to each client that it should execute profilers.command
+ def _signal_client(self, host, command):
+ """ Signal to a client that it should execute profilers.command
by writing a byte into AUTODIR/profilers.command. """
- for host in self.installed_hosts.iterkeys():
- autodir = host.get_autodir()
- path = os.path.join(autodir, "profiler.%s" % command)
- host.run("echo A > %s" % path)
+ autodir = host.get_autodir()
+ path = os.path.join(autodir, "profiler.%s" % command)
+ host.run("echo A > %s" % path)
- def start(self, test):
+ def _wait_on_client(self, host, command):
+ """ Wait for the client to signal that it's finished by writing
+ a byte into AUTODIR/profilers.command. Only waits for 30 seconds
+ before giving up. """
+ autodir = host.get_autodir()
+ path = os.path.join(autodir, "profiler.%s" % command)
+ try:
+ host.run("cat %s" % path, ignore_status=True, timeout=30)
+ except error.AutoservSSHTimeout:
+ pass # even if it times out, just give up and go ahead anyway
+
+
+ def _get_hosts(self, host=None):
+ """ Returns a dictionary of Host->Autotest mappings currently
+ supported by this profiler. If 'host' is not None, all entries
+ not matching that host object are filtered out of the dictionary."""
+ if host is None:
+ return self.installed_hosts
+ elif host in self.installed_hosts:
+ return {host: self.installed_hosts[host]}
+ else:
+ return {}
+
+
+ def start(self, test, host=None):
self._install()
encoded_args = encode_args(self.name, self.args, self.dargs)
control_script = run_profiler_control % (encoded_args, self.name)
- for at in self.installed_hosts.itervalues():
+ for host, at in self._get_hosts(host).iteritems():
+ fifo_pattern = os.path.join(host.get_autodir(), "profiler.*")
+ host.run("rm -f %s" % fifo_pattern)
at.run(control_script, background=True)
- self._signal_clients("start")
+ self._signal_client(host, "start")
+ self.current_test = test
- def stop(self, test):
- self._signal_clients("stop")
+ def stop(self, test, host=None):
+ assert self.current_test == test
+ for host in self._get_hosts(host).iterkeys():
+ self._signal_client(host, "stop")
- def report(self, test):
- self._signal_clients("report")
+ def report(self, test, host=None, wait_on_client=True):
+ assert self.current_test == test
+ self.current_test = None
+ hosts = self._get_hosts(host)
+
+ # signal to all the clients that they should report
+ if wait_on_client:
+ for host in self._get_hosts(host).iterkeys():
+ self._signal_client(host, "report")
+
# pull back all the results
- for host in self.installed_hosts.iterkeys():
+ for host in self._get_hosts(host).iterkeys():
+ if wait_on_client:
+ self._wait_on_client(host, "finished")
results_dir = os.path.join(host.get_autodir(), "results",
"default", "profiler_test",
"profiling") + "/"
local_dir = os.path.join(test.profdir, host.hostname)
if not os.path.exists(local_dir):
os.makedirs(local_dir)
+
+ tempdir = tempfile.mkdtemp(dir=self.job.tmpdir)
try:
- host.get_file(results_dir, local_dir)
+ host.get_file(results_dir, tempdir)
except error.AutoservRunError:
pass # no files to pull back, nothing we can do
+ utils.merge_trees(tempdir, local_dir)
+ shutil.rmtree(tempdir, ignore_errors=True)
+
+
+ def handle_reboot(self, host):
+ if self.current_test:
+ test = self.current_test
+ self.report(test, host, wait_on_client=False)
+ self.start(test, host)