Very rough cut of restarting a failed node
This commit is contained in:
@@ -59,54 +59,162 @@ class EachScheduling:
|
|||||||
|
|
||||||
|
|
||||||
class LoadScheduling:
|
class LoadScheduling:
|
||||||
|
"""Implement load scheduling accross nodes.
|
||||||
|
|
||||||
|
This distributes the tests collected across all nodes so each test
|
||||||
|
is run just once. All nodes collect and submit the test suit and
|
||||||
|
when all collections are received it is verified they are
|
||||||
|
identical collections. Then the collection gets devided up in
|
||||||
|
chunks and chunks get submitted to nodes. The first node
|
||||||
|
finishing it's chunk gets the next chunk until all tests are done.
|
||||||
|
|
||||||
|
When created ``numnodes`` defines how many nodes are expected to
|
||||||
|
submit a collection, this is used to know when all nodes have
|
||||||
|
finished collection or how large the chunks need to be created.
|
||||||
|
|
||||||
|
Attributes:
|
||||||
|
|
||||||
|
:numnodes: The expected number of nodes taking part. The actual
|
||||||
|
number of nodes will vary during the scheduler's lifetime as
|
||||||
|
nodes are added by the DSession as they are brought up and
|
||||||
|
removed either because of a died node or normal shutdown. This
|
||||||
|
number is primarily used to know when the initial collection is
|
||||||
|
completed.
|
||||||
|
|
||||||
|
:node2collection: Map of nodes and their test collection. All
|
||||||
|
collections should always be identical.
|
||||||
|
|
||||||
|
:node2pending: Map of nodes and the indices of their pending
|
||||||
|
tests. The indices are an index into ``.pending`` (which is
|
||||||
|
identical to their own collection stored in
|
||||||
|
``.node2collection``).
|
||||||
|
|
||||||
|
:collection: The one collection once it is validated to be
|
||||||
|
identical between all the nodes. It is initialised to None
|
||||||
|
until ``.init_distribute()`` is called.
|
||||||
|
|
||||||
|
:pending: List of indices of globally pending tests. These are
|
||||||
|
tests which have not yet been allocated to a chunk for a node
|
||||||
|
to process.
|
||||||
|
|
||||||
|
:log: A py.log.Producer instance.
|
||||||
|
|
||||||
|
"""
|
||||||
|
|
||||||
def __init__(self, numnodes, log=None):
|
def __init__(self, numnodes, log=None):
|
||||||
self.numnodes = numnodes
|
self.numnodes = numnodes
|
||||||
self.node2pending = {}
|
|
||||||
self.node2collection = {}
|
self.node2collection = {}
|
||||||
self.nodes = []
|
self.node2pending = {}
|
||||||
self.pending = []
|
self.pending = []
|
||||||
|
self.collection = None
|
||||||
if log is None:
|
if log is None:
|
||||||
self.log = py.log.Producer("loadsched")
|
self.log = py.log.Producer("loadsched")
|
||||||
else:
|
else:
|
||||||
self.log = log.loadsched
|
self.log = log.loadsched
|
||||||
self.collection_is_completed = False
|
|
||||||
|
@property
|
||||||
|
def nodes(self):
|
||||||
|
"""A list of all nodes in the scheduler."""
|
||||||
|
return list(self.node2pending.keys())
|
||||||
|
|
||||||
|
@property
|
||||||
|
def collection_is_completed(self):
|
||||||
|
"""Boolean indication initial test collection is complete.
|
||||||
|
|
||||||
|
This is a boolean indicating all initial participating nodes
|
||||||
|
have finished collection. The required number of initial
|
||||||
|
nodes is defined by ``.numnodes``.
|
||||||
|
"""
|
||||||
|
return len(self.node2collection) >= self.numnodes
|
||||||
|
|
||||||
|
def haspending(self):
|
||||||
|
"""Return True if there are pending test items
|
||||||
|
|
||||||
|
This indicates that collection has finished and nodes are
|
||||||
|
still processing test items, so can be thought of as "the
|
||||||
|
scheduler is active".
|
||||||
|
"""
|
||||||
|
if self.pending:
|
||||||
|
return True
|
||||||
|
for pending in self.node2pending.values():
|
||||||
|
if pending:
|
||||||
|
return True
|
||||||
|
return False
|
||||||
|
|
||||||
def hasnodes(self):
|
def hasnodes(self):
|
||||||
|
"""Return True if nodes exist in the scheduler."""
|
||||||
return bool(self.node2pending)
|
return bool(self.node2pending)
|
||||||
|
|
||||||
def addnode(self, node):
|
def addnode(self, node):
|
||||||
|
"""Add a new node in the scheduler.
|
||||||
|
|
||||||
|
From now on the node will be allocated chunks of tests to
|
||||||
|
execute.
|
||||||
|
|
||||||
|
Called by the ``DSession.slave_slaveready`` hook when it
|
||||||
|
sucessfully bootstrapped a new node.
|
||||||
|
"""
|
||||||
|
assert node not in self.node2pending
|
||||||
self.node2pending[node] = []
|
self.node2pending[node] = []
|
||||||
self.nodes.append(node)
|
|
||||||
|
|
||||||
def tests_finished(self):
|
def tests_finished(self):
|
||||||
|
"""Return True if all tests have been executed by the nodes."""
|
||||||
if not self.collection_is_completed:
|
if not self.collection_is_completed:
|
||||||
return False
|
return False
|
||||||
|
if self.pending:
|
||||||
|
return False
|
||||||
for pending in self.node2pending.values():
|
for pending in self.node2pending.values():
|
||||||
if len(pending) >= 2:
|
if len(pending) >= 2:
|
||||||
return False
|
return False
|
||||||
return True
|
return True
|
||||||
|
|
||||||
def addnode_collection(self, node, collection):
|
def addnode_collection(self, node, collection):
|
||||||
assert not self.collection_is_completed
|
"""Add the collected test items from a node
|
||||||
|
|
||||||
|
The collection is stored in the ``.node2collection`` map.
|
||||||
|
Called by the ``DSession.slave_collectionfinish`` hook.
|
||||||
|
"""
|
||||||
assert node in self.node2pending
|
assert node in self.node2pending
|
||||||
|
if self.collection_is_completed:
|
||||||
|
# A new node has been added later, perhaps an original one died.
|
||||||
|
assert self.collection # .init_distribute() should have
|
||||||
|
# been called by now
|
||||||
|
if collection != self.collection:
|
||||||
|
other_node = next(iter(self.node2collection.keys()))
|
||||||
|
msg = report_collection_diff(self.collection,
|
||||||
|
collection,
|
||||||
|
other_node.gateway.id,
|
||||||
|
node.gateway.id)
|
||||||
|
self.log(msg)
|
||||||
|
return
|
||||||
self.node2collection[node] = list(collection)
|
self.node2collection[node] = list(collection)
|
||||||
if len(self.node2collection) >= self.numnodes:
|
|
||||||
self.collection_is_completed = True
|
|
||||||
|
|
||||||
def remove_item(self, node, item_index, duration=0):
|
def remove_item(self, node, item_index, duration=0):
|
||||||
|
"""Mark test item as completed by node
|
||||||
|
|
||||||
|
The duration it took to execute the item is used as a hint to
|
||||||
|
the scheduler.
|
||||||
|
|
||||||
|
This is called by the ``DSession.slave_testreport`` hook.
|
||||||
|
"""
|
||||||
self.node2pending[node].remove(item_index)
|
self.node2pending[node].remove(item_index)
|
||||||
self.check_schedule(node, duration=duration)
|
self.check_schedule(node, duration=duration)
|
||||||
|
|
||||||
def check_schedule(self, node, duration=0):
|
def check_schedule(self, node, duration=0):
|
||||||
|
"""Maybe schedule new items on the node
|
||||||
|
|
||||||
|
If there are any globally pending nodes left then this will
|
||||||
|
check if the given node should be given any more tests. The
|
||||||
|
``duration`` of the last test is optionally used as a
|
||||||
|
heuristic to influence how many tests the node is assigned.
|
||||||
|
"""
|
||||||
if self.pending:
|
if self.pending:
|
||||||
# how many nodes do we have?
|
# how many nodes do we have?
|
||||||
num_nodes = len(self.node2pending)
|
num_nodes = len(self.node2pending)
|
||||||
# if our node goes below a heuristic minimum, fill it out to
|
# if our node goes below a heuristic minimum, fill it out to
|
||||||
# heuristic maximum
|
# heuristic maximum
|
||||||
items_per_node_min = max(
|
items_per_node_min = max(2, len(self.pending) // num_nodes // 4)
|
||||||
2, len(self.pending) // num_nodes // 4)
|
items_per_node_max = max(2, len(self.pending) // num_nodes // 2)
|
||||||
items_per_node_max = max(
|
|
||||||
2, len(self.pending) // num_nodes // 2)
|
|
||||||
node_pending = self.node2pending[node]
|
node_pending = self.node2pending[node]
|
||||||
if len(node_pending) < items_per_node_min:
|
if len(node_pending) < items_per_node_min:
|
||||||
if duration >= 0.1 and len(node_pending) >= 2:
|
if duration >= 0.1 and len(node_pending) >= 2:
|
||||||
@@ -116,35 +224,59 @@ class LoadScheduling:
|
|||||||
return
|
return
|
||||||
num_send = items_per_node_max - len(node_pending)
|
num_send = items_per_node_max - len(node_pending)
|
||||||
self._send_tests(node, num_send)
|
self._send_tests(node, num_send)
|
||||||
|
|
||||||
self.log("num items waiting for node:", len(self.pending))
|
self.log("num items waiting for node:", len(self.pending))
|
||||||
#self.log("node2pending:", self.node2pending)
|
|
||||||
|
|
||||||
def remove_node(self, node):
|
def remove_node(self, node):
|
||||||
self.nodes.remove(node)
|
"""Remove an node from the scheduler
|
||||||
|
|
||||||
|
This should be called either when the node crashed or at
|
||||||
|
shutdown time. In the former case any pending items assigned
|
||||||
|
to the node will be re-scheduled. Called by the
|
||||||
|
``DSession.slave_slavefinished`` and
|
||||||
|
``DSession.slave_errordown`` hooks.
|
||||||
|
|
||||||
|
Return the item which was being executing while the node
|
||||||
|
crashed or None if the node has no more pending items.
|
||||||
|
|
||||||
|
"""
|
||||||
pending = self.node2pending.pop(node)
|
pending = self.node2pending.pop(node)
|
||||||
if not pending:
|
if not pending:
|
||||||
return
|
return
|
||||||
# the node has crashed on the item if there are pending ones
|
|
||||||
# and we are told to remove the node
|
|
||||||
crashitem = self.collection[pending.pop(0)]
|
|
||||||
|
|
||||||
# put the remaining items back to the general pending list
|
# The node crashed, reassing pending items
|
||||||
|
crashitem = self.collection[pending.pop(0)]
|
||||||
self.pending.extend(pending)
|
self.pending.extend(pending)
|
||||||
# see if some nodes can pick the remaining tests up already
|
|
||||||
for node in self.node2pending:
|
for node in self.node2pending:
|
||||||
self.check_schedule(node)
|
self.check_schedule(node)
|
||||||
return crashitem
|
return crashitem
|
||||||
|
|
||||||
def init_distribute(self):
|
def init_distribute(self):
|
||||||
|
"""Initiate distribution of the test collection
|
||||||
|
|
||||||
|
Initiate scheduling of the items across the nodes. If this
|
||||||
|
gets called again later it behaves the same as calling
|
||||||
|
``.check_schedule()`` on all nodes so that newly added odes
|
||||||
|
will start to be used.
|
||||||
|
|
||||||
|
This is called by the ``DSession.slave_collectionfinish`` hook
|
||||||
|
if ``.collection_is_completed`` is True.
|
||||||
|
|
||||||
|
XXX Perhaps this method should have been called ".schedule()".
|
||||||
|
"""
|
||||||
assert self.collection_is_completed
|
assert self.collection_is_completed
|
||||||
|
|
||||||
|
# Initial distribution already happend, reschedule on all nodes
|
||||||
|
if self.collection is not None:
|
||||||
|
for node in self.nodes:
|
||||||
|
self.check_schedule(node)
|
||||||
|
return
|
||||||
|
|
||||||
# XXX allow nodes to have different collections
|
# XXX allow nodes to have different collections
|
||||||
if not self._check_nodes_have_same_collection():
|
if not self._check_nodes_have_same_collection():
|
||||||
self.log('**Different tests collected, aborting run**')
|
self.log('**Different tests collected, aborting run**')
|
||||||
return
|
return
|
||||||
|
|
||||||
# all collections are the same, good.
|
# Collections are identical, create the index of pending items.
|
||||||
# we now create an index
|
|
||||||
self.collection = list(self.node2collection.values())[0]
|
self.collection = list(self.node2collection.values())[0]
|
||||||
self.pending[:] = range(len(self.collection))
|
self.pending[:] = range(len(self.collection))
|
||||||
if not self.collection:
|
if not self.collection:
|
||||||
@@ -158,19 +290,18 @@ class LoadScheduling:
|
|||||||
for node in self.nodes:
|
for node in self.nodes:
|
||||||
self._send_tests(node, node_chunksize)
|
self._send_tests(node, node_chunksize)
|
||||||
|
|
||||||
#f = open("/tmp/sent", "w")
|
|
||||||
def _send_tests(self, node, num):
|
def _send_tests(self, node, num):
|
||||||
tests_per_node = self.pending[:num]
|
tests_per_node = self.pending[:num]
|
||||||
#print >>self.f, "sent", node, tests_per_node
|
|
||||||
if tests_per_node:
|
if tests_per_node:
|
||||||
del self.pending[:num]
|
del self.pending[:num]
|
||||||
self.node2pending[node].extend(tests_per_node)
|
self.node2pending[node].extend(tests_per_node)
|
||||||
node.send_runtest_some(tests_per_node)
|
node.send_runtest_some(tests_per_node)
|
||||||
|
|
||||||
def _check_nodes_have_same_collection(self):
|
def _check_nodes_have_same_collection(self):
|
||||||
"""
|
"""Return True if all nodes have collected the same items.
|
||||||
Return True if all nodes have collected the same items, False otherwise.
|
|
||||||
This method also logs the collection differences as they are found.
|
If collections differ this returns False and logs the
|
||||||
|
collection differences as they are found.
|
||||||
"""
|
"""
|
||||||
node_collection_items = list(self.node2collection.items())
|
node_collection_items = list(self.node2collection.items())
|
||||||
first_node, col = node_collection_items[0]
|
first_node, col = node_collection_items[0]
|
||||||
@@ -216,7 +347,20 @@ def report_collection_diff(from_collection, to_collection, from_id, to_id):
|
|||||||
class Interrupted(KeyboardInterrupt):
|
class Interrupted(KeyboardInterrupt):
|
||||||
""" signals an immediate interruption. """
|
""" signals an immediate interruption. """
|
||||||
|
|
||||||
|
|
||||||
class DSession:
|
class DSession:
|
||||||
|
"""A py.test plugin which runs a distributed test session
|
||||||
|
|
||||||
|
At the beginning of the test session this creates a NodeManager
|
||||||
|
instance which creates and starts all nodes. Nodes then emit
|
||||||
|
events processed in the pytest_runtestloop hook using the slave_*
|
||||||
|
methods.
|
||||||
|
|
||||||
|
Once a node is started it will automatically start running the
|
||||||
|
py.test mainloop with some custom hooks. This means a node
|
||||||
|
automatically starts collecting tests. Once tests are collected
|
||||||
|
it will wait for instructions.
|
||||||
|
"""
|
||||||
def __init__(self, config):
|
def __init__(self, config):
|
||||||
self.config = config
|
self.config = config
|
||||||
self.log = py.log.Producer("dsession")
|
self.log = py.log.Producer("dsession")
|
||||||
@@ -227,6 +371,7 @@ class DSession:
|
|||||||
self.maxfail = config.getvalue("maxfail")
|
self.maxfail = config.getvalue("maxfail")
|
||||||
self.queue = queue.Queue()
|
self.queue = queue.Queue()
|
||||||
self._failed_collection_errors = {}
|
self._failed_collection_errors = {}
|
||||||
|
self._active_nodes = set()
|
||||||
try:
|
try:
|
||||||
self.terminal = config.pluginmanager.getplugin("terminalreporter")
|
self.terminal = config.pluginmanager.getplugin("terminalreporter")
|
||||||
except KeyError:
|
except KeyError:
|
||||||
@@ -235,18 +380,33 @@ class DSession:
|
|||||||
self.trdist = TerminalDistReporter(config)
|
self.trdist = TerminalDistReporter(config)
|
||||||
config.pluginmanager.register(self.trdist, "terminaldistreporter")
|
config.pluginmanager.register(self.trdist, "terminaldistreporter")
|
||||||
|
|
||||||
|
@property
|
||||||
|
def session_finished(self):
|
||||||
|
"""Return True if the distributed session has finished
|
||||||
|
|
||||||
|
This means all nodes have executed all test items. This is
|
||||||
|
used to by pytest_runtestloop to break out of it's loop.
|
||||||
|
"""
|
||||||
|
return bool(self.shuttingdown and not self._active_nodes)
|
||||||
|
|
||||||
def report_line(self, line):
|
def report_line(self, line):
|
||||||
if self.terminal and self.config.option.verbose >= 0:
|
if self.terminal and self.config.option.verbose >= 0:
|
||||||
self.terminal.write_line(line)
|
self.terminal.write_line(line)
|
||||||
|
|
||||||
@pytest.mark.trylast
|
@pytest.mark.trylast
|
||||||
def pytest_sessionstart(self, session):
|
def pytest_sessionstart(self, session):
|
||||||
|
"""Creates and starts the nodes.
|
||||||
|
|
||||||
|
The nodes are setup to put their events onto self.queue. As
|
||||||
|
soon as nodes start they will emit the slave_slaveready event.
|
||||||
|
"""
|
||||||
self.nodemanager = NodeManager(self.config)
|
self.nodemanager = NodeManager(self.config)
|
||||||
self.nodemanager.setup_nodes(putevent=self.queue.put)
|
nodes = self.nodemanager.setup_nodes(putevent=self.queue.put)
|
||||||
|
self._active_nodes.update(nodes)
|
||||||
|
|
||||||
def pytest_sessionfinish(self, session):
|
def pytest_sessionfinish(self, session):
|
||||||
""" teardown any resources after a test run. """
|
"""Shutdown all nodes."""
|
||||||
nm = getattr(self, 'nodemanager', None) # if not fully initialized
|
nm = getattr(self, 'nodemanager', None) # if not fully initialized
|
||||||
if nm is not None:
|
if nm is not None:
|
||||||
nm.teardown_nodes()
|
nm.teardown_nodes()
|
||||||
|
|
||||||
@@ -264,7 +424,6 @@ class DSession:
|
|||||||
else:
|
else:
|
||||||
assert 0, dist
|
assert 0, dist
|
||||||
self.shouldstop = False
|
self.shouldstop = False
|
||||||
self.session_finished = False
|
|
||||||
while not self.session_finished:
|
while not self.session_finished:
|
||||||
self.loop_once()
|
self.loop_once()
|
||||||
if self.shouldstop:
|
if self.shouldstop:
|
||||||
@@ -272,7 +431,7 @@ class DSession:
|
|||||||
return True
|
return True
|
||||||
|
|
||||||
def loop_once(self):
|
def loop_once(self):
|
||||||
""" process one callback from one of the slaves. """
|
"""Process one callback from one of the slaves."""
|
||||||
while 1:
|
while 1:
|
||||||
try:
|
try:
|
||||||
eventcall = self.queue.get(timeout=2.0)
|
eventcall = self.queue.get(timeout=2.0)
|
||||||
@@ -293,26 +452,40 @@ class DSession:
|
|||||||
#
|
#
|
||||||
|
|
||||||
def slave_slaveready(self, node, slaveinfo):
|
def slave_slaveready(self, node, slaveinfo):
|
||||||
|
"""Emitted when a node first starts up.
|
||||||
|
|
||||||
|
This adds the node to the scheduler, nodes continue with
|
||||||
|
collection without any further input.
|
||||||
|
"""
|
||||||
node.slaveinfo = slaveinfo
|
node.slaveinfo = slaveinfo
|
||||||
node.slaveinfo['id'] = node.gateway.id
|
node.slaveinfo['id'] = node.gateway.id
|
||||||
node.slaveinfo['spec'] = node.gateway.spec
|
node.slaveinfo['spec'] = node.gateway.spec
|
||||||
self.config.hook.pytest_testnodeready(node=node)
|
self.config.hook.pytest_testnodeready(node=node)
|
||||||
self.sched.addnode(node)
|
|
||||||
if self.shuttingdown:
|
if self.shuttingdown:
|
||||||
node.shutdown()
|
node.shutdown()
|
||||||
|
else:
|
||||||
|
self.sched.addnode(node)
|
||||||
|
|
||||||
def slave_slavefinished(self, node):
|
def slave_slavefinished(self, node):
|
||||||
|
"""Emitted when node executes its pytest_sessionfinish hook.
|
||||||
|
|
||||||
|
Removes the node from the scheduler.
|
||||||
|
|
||||||
|
The node might not be the scheduler if it had not emitted
|
||||||
|
slaveready before shutdown was triggered.
|
||||||
|
"""
|
||||||
self.config.hook.pytest_testnodedown(node=node, error=None)
|
self.config.hook.pytest_testnodedown(node=node, error=None)
|
||||||
if node.slaveoutput['exitstatus'] == 2: # keyboard-interrupt
|
if node.slaveoutput['exitstatus'] == 2: # keyboard-interrupt
|
||||||
self.shouldstop = "%s received keyboard-interrupt" % (node,)
|
self.shouldstop = "%s received keyboard-interrupt" % (node,)
|
||||||
self.slave_errordown(node, "keyboard-interrupt")
|
self.slave_errordown(node, "keyboard-interrupt")
|
||||||
return
|
return
|
||||||
crashitem = self.sched.remove_node(node)
|
if node in self.sched.nodes:
|
||||||
assert not crashitem, (crashitem, node)
|
crashitem = self.sched.remove_node(node)
|
||||||
if self.shuttingdown and not self.sched.hasnodes():
|
assert not crashitem, (crashitem, node)
|
||||||
self.session_finished = True
|
self._active_nodes.remove(node)
|
||||||
|
|
||||||
def slave_errordown(self, node, error):
|
def slave_errordown(self, node, error):
|
||||||
|
"""Emitted by the SlaveController when a node dies."""
|
||||||
self.config.hook.pytest_testnodedown(node=node, error=error)
|
self.config.hook.pytest_testnodedown(node=node, error=error)
|
||||||
try:
|
try:
|
||||||
crashitem = self.sched.remove_node(node)
|
crashitem = self.sched.remove_node(node)
|
||||||
@@ -321,41 +494,70 @@ class DSession:
|
|||||||
else:
|
else:
|
||||||
if crashitem:
|
if crashitem:
|
||||||
self.handle_crashitem(crashitem, node)
|
self.handle_crashitem(crashitem, node)
|
||||||
#self.report_line("item crashed on node: %s" % crashitem)
|
self.report_line("Replacing failed node %s" % node.gateway.id)
|
||||||
if not self.sched.hasnodes():
|
self._clone_node(node)
|
||||||
self.session_finished = True
|
self._active_nodes.remove(node)
|
||||||
|
|
||||||
def slave_collectionfinish(self, node, ids):
|
def slave_collectionfinish(self, node, ids):
|
||||||
|
"""Slave has finished test collection.
|
||||||
|
|
||||||
|
This adds the collection for this node to the scheduler. If
|
||||||
|
the scheduler indicates collection is finished (i.e. all
|
||||||
|
initial nodes have submitted their collection), then tells the
|
||||||
|
scheduler to schedule the collected items. When initiating
|
||||||
|
scheduling the first time it logs which scheduler is in use.
|
||||||
|
"""
|
||||||
|
if self.shuttingdown:
|
||||||
|
return
|
||||||
self.sched.addnode_collection(node, ids)
|
self.sched.addnode_collection(node, ids)
|
||||||
if self.terminal:
|
if self.terminal:
|
||||||
self.trdist.setstatus(node.gateway.spec, "[%d]" %(len(ids)))
|
self.trdist.setstatus(node.gateway.spec, "[%d]" % (len(ids)))
|
||||||
|
|
||||||
if self.sched.collection_is_completed:
|
if self.sched.collection_is_completed:
|
||||||
if self.terminal:
|
if self.terminal and not self.sched.haspending():
|
||||||
self.trdist.ensure_show_status()
|
self.trdist.ensure_show_status()
|
||||||
self.terminal.write_line("")
|
self.terminal.write_line("")
|
||||||
self.terminal.write_line("scheduling tests via %s" %(
|
self.terminal.write_line("scheduling tests via %s" % (
|
||||||
self.sched.__class__.__name__))
|
self.sched.__class__.__name__))
|
||||||
|
|
||||||
self.sched.init_distribute()
|
self.sched.init_distribute()
|
||||||
|
|
||||||
def slave_logstart(self, node, nodeid, location):
|
def slave_logstart(self, node, nodeid, location):
|
||||||
|
"""Emitted when a node calls the pytest_runtest_logstart hook."""
|
||||||
self.config.hook.pytest_runtest_logstart(
|
self.config.hook.pytest_runtest_logstart(
|
||||||
nodeid=nodeid, location=location)
|
nodeid=nodeid, location=location)
|
||||||
|
|
||||||
def slave_testreport(self, node, rep):
|
def slave_testreport(self, node, rep):
|
||||||
if not (rep.passed and rep.when != "call"):
|
"""Emitted when a node calls the pytest_runtest_logreport hook.
|
||||||
if rep.when in ("setup", "call"):
|
|
||||||
self.sched.remove_item(node, rep.item_index, rep.duration)
|
If the node indicates it is finished with a test item remove
|
||||||
|
the item from the pending list in the scheduler.
|
||||||
|
"""
|
||||||
|
if rep.when == "call" or (rep.when == "setup" and not rep.passed):
|
||||||
|
self.sched.remove_item(node, rep.item_index, rep.duration)
|
||||||
#self.report_line("testreport %s: %s" %(rep.id, rep.status))
|
#self.report_line("testreport %s: %s" %(rep.id, rep.status))
|
||||||
rep.node = node
|
rep.node = node
|
||||||
self.config.hook.pytest_runtest_logreport(report=rep)
|
self.config.hook.pytest_runtest_logreport(report=rep)
|
||||||
self._handlefailures(rep)
|
self._handlefailures(rep)
|
||||||
|
|
||||||
def slave_collectreport(self, node, rep):
|
def slave_collectreport(self, node, rep):
|
||||||
|
"""Emitted when a node calls the pytest_collectreport hook."""
|
||||||
if rep.failed:
|
if rep.failed:
|
||||||
self._failed_slave_collectreport(node, rep)
|
self._failed_slave_collectreport(node, rep)
|
||||||
|
|
||||||
|
def _clone_node(self, node):
|
||||||
|
"""Return new node based on an existing one.
|
||||||
|
|
||||||
|
This is normally for when a node died, this will copy the spec
|
||||||
|
of the existing node and create a new one with a new id. The
|
||||||
|
new node will have been setup so will start calling the
|
||||||
|
"slave_*" hooks and do work soon.
|
||||||
|
"""
|
||||||
|
spec = node.gateway.spec
|
||||||
|
spec.id = None
|
||||||
|
self.nodemanager.group.allocate_id(spec)
|
||||||
|
node = self.nodemanager.setup_node(spec, self.queue.put)
|
||||||
|
self._active_nodes.add(node)
|
||||||
|
return node
|
||||||
|
|
||||||
def _failed_slave_collectreport(self, node, rep):
|
def _failed_slave_collectreport(self, node, rep):
|
||||||
# Check we haven't already seen this report (from
|
# Check we haven't already seen this report (from
|
||||||
# another slave).
|
# another slave).
|
||||||
@@ -374,19 +576,21 @@ class DSession:
|
|||||||
def triggershutdown(self):
|
def triggershutdown(self):
|
||||||
self.log("triggering shutdown")
|
self.log("triggering shutdown")
|
||||||
self.shuttingdown = True
|
self.shuttingdown = True
|
||||||
for node in self.sched.node2pending:
|
for node in self.sched.nodes:
|
||||||
node.shutdown()
|
node.shutdown()
|
||||||
|
|
||||||
def handle_crashitem(self, nodeid, slave):
|
def handle_crashitem(self, nodeid, slave):
|
||||||
# XXX get more reporting info by recording pytest_runtest_logstart?
|
# XXX get more reporting info by recording pytest_runtest_logstart?
|
||||||
|
# XXX count no of failures and retry N times
|
||||||
runner = self.config.pluginmanager.getplugin("runner")
|
runner = self.config.pluginmanager.getplugin("runner")
|
||||||
fspath = nodeid.split("::")[0]
|
fspath = nodeid.split("::")[0]
|
||||||
msg = "Slave %r crashed while running %r" %(slave.gateway.id, nodeid)
|
msg = "Slave %r crashed while running %r" % (slave.gateway.id, nodeid)
|
||||||
rep = runner.TestReport(nodeid, (fspath, None, fspath), (),
|
rep = runner.TestReport(nodeid, (fspath, None, fspath),
|
||||||
"failed", msg, "???")
|
(), "failed", msg, "???")
|
||||||
rep.node = slave
|
rep.node = slave
|
||||||
self.config.hook.pytest_runtest_logreport(report=rep)
|
self.config.hook.pytest_runtest_logreport(report=rep)
|
||||||
|
|
||||||
|
|
||||||
class TerminalDistReporter:
|
class TerminalDistReporter:
|
||||||
def __init__(self, config):
|
def __init__(self, config):
|
||||||
self.config = config
|
self.config = config
|
||||||
|
|||||||
@@ -28,34 +28,32 @@ class NodeManager(object):
|
|||||||
self.specs.append(spec)
|
self.specs.append(spec)
|
||||||
self.roots = self._getrsyncdirs()
|
self.roots = self._getrsyncdirs()
|
||||||
self.rsyncoptions = self._getrsyncoptions()
|
self.rsyncoptions = self._getrsyncoptions()
|
||||||
|
self._rsynced_specs = py.builtin.set()
|
||||||
|
|
||||||
def rsync_roots(self):
|
def rsync_roots(self, gateway):
|
||||||
""" make sure that all remote gateways
|
"""Rsync the set of roots to the node's gateway cwd."""
|
||||||
have the same set of roots in their
|
|
||||||
current directory.
|
|
||||||
"""
|
|
||||||
if self.roots:
|
if self.roots:
|
||||||
# send each rsync root
|
|
||||||
for root in self.roots:
|
for root in self.roots:
|
||||||
self.rsync(root, **self.rsyncoptions)
|
self.rsync(gateway, root, **self.rsyncoptions)
|
||||||
|
|
||||||
def makegateways(self):
|
|
||||||
assert not list(self.group)
|
|
||||||
self.config.hook.pytest_xdist_setupnodes(config=self.config,
|
|
||||||
specs=self.specs)
|
|
||||||
for spec in self.specs:
|
|
||||||
gw = self.group.makegateway(spec)
|
|
||||||
self.config.hook.pytest_xdist_newgateway(gateway=gw)
|
|
||||||
|
|
||||||
def setup_nodes(self, putevent):
|
def setup_nodes(self, putevent):
|
||||||
self.makegateways()
|
self.config.hook.pytest_xdist_setupnodes(config=self.config,
|
||||||
self.rsync_roots()
|
specs=self.specs)
|
||||||
self.trace("setting up nodes")
|
self.trace("setting up nodes")
|
||||||
for gateway in self.group:
|
nodes = []
|
||||||
node = SlaveController(self, gateway, self.config, putevent)
|
for spec in self.specs:
|
||||||
gateway.node = node # to keep node alive
|
nodes.append(self.setup_node(spec, putevent))
|
||||||
node.setup()
|
return nodes
|
||||||
self.trace("started node %r" % node)
|
|
||||||
|
def setup_node(self, spec, putevent):
|
||||||
|
gw = self.group.makegateway(spec)
|
||||||
|
self.config.hook.pytest_xdist_newgateway(gateway=gw)
|
||||||
|
self.rsync_roots(gw)
|
||||||
|
node = SlaveController(self, gw, self.config, putevent)
|
||||||
|
gw.node = node # keep the node alive
|
||||||
|
node.setup()
|
||||||
|
self.trace("started node %r" % node)
|
||||||
|
return node
|
||||||
|
|
||||||
def teardown_nodes(self):
|
def teardown_nodes(self):
|
||||||
self.group.terminate(self.EXIT_TIMEOUT)
|
self.group.terminate(self.EXIT_TIMEOUT)
|
||||||
@@ -110,39 +108,38 @@ class NodeManager(object):
|
|||||||
'verbose': self.config.option.verbose,
|
'verbose': self.config.option.verbose,
|
||||||
}
|
}
|
||||||
|
|
||||||
|
def rsync(self, gateway, source, notify=None, verbose=False, ignores=None):
|
||||||
def rsync(self, source, notify=None, verbose=False, ignores=None):
|
"""Perform rsync to remote hosts for node."""
|
||||||
""" perform rsync to all remote hosts.
|
# XXX Probably want to keep a list of rsynced specs to avoid
|
||||||
"""
|
# duplicate rsyncs.
|
||||||
|
# XXX This changes the calling behaviour of
|
||||||
|
# pytest_xdist_rsyncstart and pytest_xdist_rsyncfinish to
|
||||||
|
# be called once per rsync target.
|
||||||
rsync = HostRSync(source, verbose=verbose, ignores=ignores)
|
rsync = HostRSync(source, verbose=verbose, ignores=ignores)
|
||||||
seen = py.builtin.set()
|
spec = gateway.spec
|
||||||
gateways = []
|
if spec.popen and not spec.chdir:
|
||||||
for gateway in self.group:
|
# XXX This assumes that sources are python-packages
|
||||||
spec = gateway.spec
|
# and that adding the basedir does not hurt.
|
||||||
if spec.popen and not spec.chdir:
|
gateway.remote_exec("""
|
||||||
# XXX this assumes that sources are python-packages
|
import sys ; sys.path.insert(0, %r)
|
||||||
# and that adding the basedir does not hurt
|
""" % os.path.dirname(str(source))).waitclose()
|
||||||
gateway.remote_exec("""
|
return
|
||||||
import sys ; sys.path.insert(0, %r)
|
if spec in self._rsynced_specs:
|
||||||
""" % os.path.dirname(str(source))).waitclose()
|
return
|
||||||
continue
|
def finished():
|
||||||
if spec not in seen:
|
if notify:
|
||||||
def finished():
|
notify("rsyncrootready", spec, source)
|
||||||
if notify:
|
rsync.add_target_host(gateway, finished=finished)
|
||||||
notify("rsyncrootready", spec, source)
|
self._rsynced_specs.add(spec)
|
||||||
rsync.add_target_host(gateway, finished=finished)
|
self.config.hook.pytest_xdist_rsyncstart(
|
||||||
seen.add(spec)
|
source=source,
|
||||||
gateways.append(gateway)
|
gateways=[gateway],
|
||||||
if seen:
|
)
|
||||||
self.config.hook.pytest_xdist_rsyncstart(
|
rsync.send()
|
||||||
source=source,
|
self.config.hook.pytest_xdist_rsyncfinish(
|
||||||
gateways=gateways,
|
source=source,
|
||||||
)
|
gateways=[gateway],
|
||||||
rsync.send()
|
)
|
||||||
self.config.hook.pytest_xdist_rsyncfinish(
|
|
||||||
source=source,
|
|
||||||
gateways=gateways,
|
|
||||||
)
|
|
||||||
|
|
||||||
class HostRSync(execnet.RSync):
|
class HostRSync(execnet.RSync):
|
||||||
""" RSyncer that filters out common files
|
""" RSyncer that filters out common files
|
||||||
|
|||||||
Reference in New Issue
Block a user