Yes, I confirmed this works on your watcher rework branch. (Nothing I was able to hack on CURATOR-3.0 would make it pass, probably due to collection element equality/hashcode issues with wrapping in a spot I wasn't able to find.)
On Fri, Feb 5, 2016 at 11:31 AM, Scott Blum <[email protected]> wrote: > Ahh.. super slick. So recipes no longer have to remove watches on deleted > nodes, the framework does it for you. > > On Fri, Feb 5, 2016 at 9:46 AM, Jordan Zimmerman < > [email protected]> wrote: > >> The error is occurring because of this line in wasDeleted(): >> >> >> client.watches().remove(this).ofType(WatcherType.Any).locally().inBackground().forPath(path); >> >> I believe this is unnecessary as the WatcherRemoveCuratorFramework >> handles it automatically. If I remove this line the log errors go away. >> >> -Jordan >> >> On Feb 3, 2016, at 7:45 PM, Scott Blum <[email protected]> wrote: >> >> TestTreeCache.testBasics() fails during teardown on CURATOR-3.0 (4e4072b) >> >> On Wed, Feb 3, 2016 at 7:21 PM, Jordan Zimmerman < >> [email protected]> wrote: >> >>> Scott, do you have a small test that shows this error? I’d really like >>> to fix it and can’t get a reproducible case. >>> >>> -Jordan >>> >>> On Feb 1, 2016, at 3:14 PM, Scott Blum <[email protected]> wrote: >>> >>> Alright, I've git bisected it down to a single change that breaks >>> TestTreeCache: >>> >>> WatcherRemovalManager.internalRemove, >>> >>> private synchronized void internalRemove(WrappedWatcher entry) >>> { >>> namespaceWatcherMap.removeWatcher(entry.watcher); >>> entries.remove(entry); >>> } >>> >>> The offending line seems to be the namespaceWatcherMap.removeWatcher. >>> If I comment out that line, TestTreeCache passes. >>> >>> >>> On Mon, Feb 1, 2016 at 3:05 PM, Scott Blum <[email protected]> >>> wrote: >>> >>>> I think there may be some general problem, I notice warning messages in >>>> the test logs, like: >>>> >>>> ERROR org.apache.zookeeper.ClientCnxn Failed to find watcher! >>>> [main-SendThread(127.0.0.1:60086)] >>>> org.apache.zookeeper.KeeperException$NoWatcherException: >>>> KeeperErrorCode = No such watcher for /test/one >>>> at >>>> org.apache.zookeeper.ZooKeeper$ZKWatchManager.containsWatcher(ZooKeeper.java:377) >>>> at >>>> org.apache.zookeeper.ZooKeeper$ZKWatchManager.removeWatcher(ZooKeeper.java:252) >>>> at >>>> org.apache.zookeeper.WatchDeregistration.unregister(WatchDeregistration.java:58) >>>> >>>> >>>> On Mon, Feb 1, 2016 at 3:04 PM, Scott Blum <[email protected]> >>>> wrote: >>>> >>>>> Actually, it's not a problem with the changes I'm trying to merge in; >>>>> TestTreeCache is currently broken on 3.0 branch already. >>>>> >>>>> Looks like it broke on >>>>> revision c6a22ba508f9227fe1c657f93e3cc77d8bc17e3e, which was a merge from >>>>> master into 3.0, with some conflict resolution. >>>>> >>>>> On Mon, Feb 1, 2016 at 2:46 PM, Jordan Zimmerman < >>>>> [email protected]> wrote: >>>>> >>>>>> I can try to merge when I get a chance. At some point, we should move >>>>>> CURATOR-3.0 to master so we don’t have to keep maintaining this. >>>>>> >>>>>> -JZ >>>>>> >>>>>> On Feb 1, 2016, at 2:38 PM, Scott Blum <[email protected]> wrote: >>>>>> >>>>>> TestTreeCache fails on 6acf0987125469ecb79ff48ce5e9735a98cd1abb >>>>>> <http://git-wip-us.apache.org/repos/asf/curator/commit/6acf0987>, a >>>>>> proposed merge of master into 3.0. >>>>>> >>>>>> It has something to do with the new watcher stuff, but I haven't been >>>>>> able to figure out what the problem is, possible because I don't >>>>>> understand >>>>>> the new stuff well enough. Any ideas? >>>>>> >>>>>> >>>>>> >>>>> >>>> >>> >>> >> >> >
