Tests for select-specific failure conditions.
Method | port |
Create, start, and return a new Port , also tracking it so it can be stopped in the test tear down. |
Method | set |
Undocumented |
Method | tear |
Undocumented |
Method | test_accept |
tcp.Port.doRead increases the number of consecutive accept calls it performs if all of the previous accept calls succeed; otherwise, it reduces the number to the amount of successful calls. |
Method | test_connection |
Similar to test_tooManyFilesFromAccept , but test the case where accept(2) fails with ECONNABORTED. |
Method | test_no |
Similar to test_tooManyFilesFromAccept , but test the case where accept(2) fails with ENOBUFS. |
Method | test_no |
Similar to test_tooManyFilesFromAccept , but test the case where accept(2) fails with ENFILE. |
Method | test_no |
Similar to test_tooManyFilesFromAccept , but test the case where accept(2) fails with ENOMEM. |
Method | test_permission |
accept(2) returning EPERM is treated as a transient failure and the call retried no more than the maximum number of consecutive accept(2) calls. |
Method | test_too |
accept(2) can fail with EMFILE when there are too many open file descriptors in the process. Test that this doesn't negatively impact any other existing connections. |
Method | test_unknown |
A socket.error raised by accept(2) whose errno is unknown to the recovery logic is logged. |
Instance Variable | messages |
Undocumented |
Instance Variable | ports |
Undocumented |
Method | _accept |
Test behavior in the face of an exception from accept(2). |
Inherited from TestCase
:
Method | __call__ |
Run the test. Should always do exactly the same thing as run(). |
Method | __init__ |
Construct an asynchronous test case for methodName. |
Method | add |
Extend the base cleanup feature with support for cleanup functions which return Deferreds. |
Method | assert |
Fail if deferred does not errback with one of expectedFailures. Returns the original Deferred with callbacks added. You will need to return this Deferred from your test case. |
Method | defer |
Run any scheduled cleanups and report errors (if any) to the result. object. |
Method | defer |
Undocumented |
Method | defer |
Undocumented |
Method | defer |
Undocumented |
Method | get |
Undocumented |
Method | get |
Returns the timeout value set on this test. Checks on the instance first, then the class, then the module, then packages. As soon as it finds something with a timeout attribute, returns that. Returns util.DEFAULT_TIMEOUT_DURATION ... |
Instance Variable | timeout |
A real number of seconds. If set, the test will raise an error if it takes longer than timeout seconds. If not set, util.DEFAULT_TIMEOUT_DURATION is used. |
Method | _cb |
Undocumented |
Method | _class |
Undocumented |
Method | _clean |
Undocumented |
Method | _deprecate |
Deprecate iterate, crash and stop on reactor. That is, each method is wrapped in a function that issues a deprecation warning, then calls the original. |
Method | _eb |
Undocumented |
Method | _eb |
Undocumented |
Method | _eb |
Undocumented |
Method | _make |
Create a method which wraps the reactor method name. The new method issues a deprecation warning and calls the original. |
Method | _run |
Run a single method, either a test method or fixture. |
Method | _run |
Really run setUp, the test method, and tearDown. Any of these may return defer.Deferred s. After they complete, do some reactor cleanup. |
Method | _undeprecate |
Restore the deprecated reactor methods. Undoes what _deprecateReactor did. |
Method | _wait |
Take a Deferred that only ever callbacks. Block until it happens. |
Instance Variable | _passed |
Undocumented |
Instance Variable | _reactor |
Undocumented |
Instance Variable | _timed |
Undocumented |
Inherited from SynchronousTestCase
(via TestCase
):
Method | __eq__ |
Override the comparison defined by the base TestCase which considers instances of the same class with the same _testMethodName to be equal. Since trial puts TestCase instances into a set, that definition of comparison makes it impossible to run the same test method twice... |
Method | __hash__ |
Undocumented |
Method | call |
Call a function that should have been deprecated at a specific version and in favor of a specific alternative, and assert that it was thusly deprecated. |
Method | flush |
Remove stored errors received from the log. |
Method | flush |
Remove stored warnings from the list of captured warnings and return them. |
Method | get |
Retrieve a module attribute which should have been deprecated, and assert that we saw the appropriate deprecation warning. |
Method | get |
Return the skip reason set on this test, if any is set. Checks on the instance first, then the class, then the module, then packages. As soon as it finds something with a skip attribute, returns that in a tuple (... |
Method | get |
Return a Todo object if the test is marked todo. Checks on the instance first, then the class, then the module, then packages. As soon as it finds something with a todo attribute, returns that. Returns ... |
Method | mktemp |
Create a new path name which can be used for a new file or directory. |
Method | patch |
Monkey patch an object for the duration of the test. |
Method | run |
Run the test case, storing the results in result. |
Method | run |
If no methodName argument is passed to the constructor, run will treat this method as the thing with the actual test inside. |
Method | short |
Undocumented |
Instance Variable | failure |
An exception class, defaulting to FailTest. If the test method raises this exception, it will be reported as a failure, rather than an exception. All of the assertion methods raise this if the assertion fails. |
Instance Variable | skip |
None or a string explaining why this test is to be skipped. If defined, the test will not be run. Instead, it will be reported to the result object as 'skipped' (if the TestResult supports skipping). |
Instance Variable | suppress |
None or a list of tuples of (args, kwargs) to be passed to warnings.filterwarnings. Use these to suppress warnings raised in a test. Useful for testing deprecated code. See also util.suppress . |
Instance Variable | todo |
None , a string or a tuple of (errors, reason) where errors is either an exception class or an iterable of exception classes, and reason is a string. See Todo or makeTodo for more information. |
Method | _get |
Return the reason to use for skipping a test method. |
Method | _get |
Returns any warning suppressions set for this test. Checks on the instance first, then the class, then the module, then packages. As soon as it finds something with a suppress attribute, returns that. ... |
Method | _install |
Undocumented |
Method | _remove |
Undocumented |
Method | _run |
Synchronously run any cleanups which have been added. |
Instance Variable | _cleanups |
Undocumented |
Instance Variable | _observer |
Undocumented |
Instance Variable | _parents |
Undocumented |
Instance Variable | _test |
Undocumented |
Instance Variable | _warnings |
Undocumented |
Inherited from _Assertions
(via TestCase
, SynchronousTestCase
):
Method | assert |
Fail if the two objects are unequal as determined by their difference rounded to the given number of decimal places (default 7) and comparing to zero. |
Method | assert |
Fail if first - second > tolerance |
Method | assert |
Fail the test if first and second are not equal. |
Method | assert |
Fail the test if condition evaluates to True. |
Method | assert |
Fail the test if containee is not found in container. |
Method | assert |
Fail the test if first is not second. This is an obect-identity-equality test, not an object equality (i.e. __eq__) test. |
Method | assert |
Fail if instance is not an instance of the given class or of one of the given classes. |
Method | assert |
Fail the test if first is second. This is an obect-identity-equality test, not an object equality (i.e. __eq__) test. |
Method | assert |
Assert that deferred does not have a result at this point. |
Method | assert |
Fail if the two objects are equal as determined by their difference rounded to the given number of decimal places (default 7) and comparing to zero. |
Method | assert |
Fail the test if first == second. |
Method | assert |
Fail the test if containee is found in container. |
Method | assert |
Fail if instance is an instance of the given class or of one of the given classes. |
Method | assert |
Fail if astring contains substring. |
Method | assert |
Fail the test unless calling the function f with the given args and kwargs raises exception. The failure will report the traceback and call stack of the unexpected exception. |
Method | assert |
Fail if substring does not exist within astring. |
Method | assert |
Fail the test if condition evaluates to False. |
Method | assert |
Fail if the given function doesn't generate the specified warning when called. It calls the function, checks the warning, and forwards the result of the function if everything is fine. |
Method | fail |
Absolutely fail the test. Do not pass go, do not collect $200. |
Method | failure |
Return the current failure result of deferred or raise self.failureException. |
Method | success |
Return the current success result of deferred or raise self.failureException. |
def test_acceptScaling(self): (source) ¶
tcp.Port.doRead
increases the number of consecutive accept calls it performs if all of the previous accept calls succeed; otherwise, it reduces the number to the amount of successful calls.
Similar to test_tooManyFilesFromAccept
, but test the case where accept(2) fails with ECONNABORTED.
It is not clear whether this is actually possible for TCP connections on modern versions of Linux.
Similar to test_tooManyFilesFromAccept
, but test the case where accept(2) fails with ENOBUFS.
This mainly occurs on Windows and FreeBSD, but may be possible on Linux and other platforms as well.
def test_noFilesFromAccept(self): (source) ¶
Similar to test_tooManyFilesFromAccept
, but test the case where accept(2) fails with ENFILE.
This can occur on Linux when the system has exhausted (!) its supply of inodes.
def test_noMemoryFromAccept(self): (source) ¶
Similar to test_tooManyFilesFromAccept
, but test the case where accept(2) fails with ENOMEM.
On Linux at least, this can sensibly occur, even in a Python program (which eats memory like no ones business), when memory has become fragmented or low memory has been filled (d_alloc calls kmem_cache_alloc calls kmalloc - kmalloc only allocates out of low memory).
def test_permissionFailure(self): (source) ¶
accept(2) returning EPERM is treated as a transient failure and the call retried no more than the maximum number of consecutive accept(2) calls.
accept(2) can fail with EMFILE when there are too many open file descriptors in the process. Test that this doesn't negatively impact any other existing connections.
EMFILE mainly occurs on Linux when the open file rlimit is encountered.
Test behavior in the face of an exception from accept(2).
On any exception which indicates the platform is unable or unwilling to allocate further resources to us, the existing port should remain listening, a message should be logged, and the exception should not propagate outward from doRead.
Parameters | |
socket | The errno to simulate from accept. |