Dan Kenigsberg has posted comments on this change. Change subject: executor: introduce the executor library ......................................................................
Patch Set 31: (3 comments) http://gerrit.ovirt.org/#/c/29191/31/lib/vdsm/executor.py File lib/vdsm/executor.py: Line 144: self._scheduler = scheduler Line 145: self._busy = False Line 146: self._discarded = False Line 147: _Worker._id += 1 Line 148: name = "Executor-%i" % _Worker._id > Not sure what you sugget, but letting the executor set the name of the work I was suggesting "executor-m-worker-n"; since multiple executors is not a big concern for us right now, I can do with "worker-n". "executor-n", as suggested in the code, is a confusing name for a worker. Line 149: self._thread = threading.Thread(target=self._run, name=name) Line 150: self._thread.daemon = True Line 151: self._log.debug('Starting worker %s' % name) Line 152: self._thread.start() Line 154: @property Line 155: def name(self): Line 156: return self._thread.name Line 157: Line 158: def join(self): > We join when Executor.stop() is invoke with wait == True. This is the defau usually, join() is called when you WANT to wait for a certain job to be done. That's pretty useful some times. I cannot find a use case for waiting only for threads that are doing nothing. Line 159: if self._busy: Line 160: raise _WorkerBusy() Line 161: self._log.debug('Waiting for worker %s', self.name) Line 162: self._thread.join() Line 231: except IndexError: Line 232: with self._cond: Line 233: self._waiters += 1 Line 234: if not self._tasks: Line 235: self._cond.wait() > The spurious wakeup bug is not related to multiple threads, but to wait ret I forgot that wait() reacquires the mutex once woken up. There's no race here. Can you explain why we need self._waiters? What if put() is called right before _waiters += 1? put() would skip the notify, and get() would end up waiting forever. Line 236: self._waiters -= 1 Line 237: Line 238: def clear(self): -- To view, visit http://gerrit.ovirt.org/29191 To unsubscribe, visit http://gerrit.ovirt.org/settings Gerrit-MessageType: comment Gerrit-Change-Id: Ic06da1ba57868dc2c7db67a1868ad10087a1cff2 Gerrit-PatchSet: 31 Gerrit-Project: vdsm Gerrit-Branch: master Gerrit-Owner: Francesco Romani <[email protected]> Gerrit-Reviewer: Antoni Segura Puimedon <[email protected]> Gerrit-Reviewer: Dan Kenigsberg <[email protected]> Gerrit-Reviewer: Dima Kuznetsov <[email protected]> Gerrit-Reviewer: Federico Simoncelli <[email protected]> Gerrit-Reviewer: Francesco Romani <[email protected]> Gerrit-Reviewer: Michal Skrivanek <[email protected]> Gerrit-Reviewer: Nir Soffer <[email protected]> Gerrit-Reviewer: Saggi Mizrahi <[email protected]> Gerrit-Reviewer: Vinzenz Feenstra <[email protected]> Gerrit-Reviewer: Yaniv Bronhaim <[email protected]> Gerrit-Reviewer: [email protected] Gerrit-Reviewer: oVirt Jenkins CI Server Gerrit-HasComments: Yes _______________________________________________ vdsm-patches mailing list [email protected] https://lists.fedorahosted.org/mailman/listinfo/vdsm-patches
