| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Closes #255
As of Python 3.5, PEP 475 gets in our way. Runners with long time.sleep()'s
in their _snooze() method (e.g. the retry runner) will have their system call
implemented time.sleep() automatically retried at the C layer. The only
reliable way to prevent this is to raise an exception in the signal handler.
The standard run() method automatically suppresses this exception, meaning,
it's caught and ignored, but effectively breaks the run() loop, which is just
what we want.
The lmtp and rest runners implement their own run loops, so they also have to
handle this exception, by ignoring it.
|
| | |
|
| | |
|
| |
|
|
| |
Fix lots of import order errors discovered by the new plugin.
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
|
|
| |
reflect this.
Also, mock out sys.stderr on some tests so that their nose2 output is quieter.
A few other minor coding style consistencies.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
reduced. Now, only one `-r/--runner` option may be provided and the
round-robin feature has been removed.
* Fixed REST server crash on `reopen` command. Identification and test
provided by Aurélien Bompard. (LP: #1184376)
Also:
* bin/runner now uses standard argparse instead of ScriptOptions.
* The entire bin/runner machinery has bee reorganized and simplified. There
* is no more Loop class. Signal setting is moved directly into the base
Runner class and overrided in specific subclasses (e.g. RESTRunner which
must cleanly shutdown its TCPServer). The runner exit status is now set
directly on the Runner instance.
* Fixed a few minor style issues.
* In order to cleanly shutdown the RESTRunner's WSGI server, we must start a
subthread which only watches for an Event and then calls the server's
shutdown() method. It has to be this way because the WSGI server itself
(due to interactions with SQLite), and the signal handlers (due to Python's
signal handling semantics) must both run in the main thread. However, the
shutdown() must be invoked from a subthread in order to prevent deadlock.
* Refactor the RESTLayer to eliminate duplication of code.
|
| | |
|
| |\
| |
| |
| | |
Sandesh Kumar Agrawal. (LP: #1095422)
|
| |/ |
|
| | |
|
| |
|
|
|
|
| |
`RunnerCrashEvent` is triggered, which contains references to the queue
runner, mailing list, message, metadata, and exception. Interested parties
can subscribe to that `zope.event` for notification.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* Queue runners are now called just 'Runners' since several of them don't
manage queue directories.
* Ban the term 'qrunner' too.
* The master queue runner watcher should now just be called the 'master' or
the 'master runner'.
* bin/qrunner -> bin/runner
* mailman.qrunner log file -> mailman.runner
* master-qrunner.lck -> master.lck
* master-qrunner.pid -> master.pid
Also:
* Remove some obsolete files
* Begin the .txt -> .rst renaming
|
| | |
|
| | |
|
| |
|
|
|
| |
runner flag to indicate whether the qrunner parent process should intercept
signals or not. The REST server should not intercept signals.
|
|
|
correctly regardless of how it's used.
|