Logged under https://issues.apache.org/jira/browse/CURATOR-391
On Tue, Mar 7, 2017 at 10:45 AM, Jordan Zimmerman < [email protected]> wrote: > Oh - interesting. If you don't mind - please open an issue in our Jira. > This is a bug. > > -Jordan > > On Mar 7, 2017, at 10:44 AM, Benjamin Jaton <[email protected]> > wrote: > > if ( previousData.getStat().getVersion() != stat.getVersion() ) > > I think this is the problem. > If the node is deleted then recreated, the version will be reset to 0. > > So for instance in my case, the previous stat data version is 0. > Then I recreate the node with different data, but the version is also 0. > > > On Mon, Mar 6, 2017 at 12:17 PM, Jordan Zimmerman < > [email protected]> wrote: > >> PathChildrenCache sets a ConnectionStateListener. When the RECONNECTED >> state is received, it refreshes the cache. You will only get a >> CHILD_UPDATED if the Stat Version has changed. See the applyNewData() >> method. >> >> -Jordan >> >> > On Mar 6, 2017, at 12:05 PM, Benjamin Jaton <[email protected]> >> wrote: >> > >> > I am getting those events using PathChildrenCacheListener: >> > >> > PathChildrenCacheEvent{type=CONNECTION_SUSPENDED, data=null} >> > PathChildrenCacheEvent{type=CONNECTION_LOST, data=null} >> > PathChildrenCacheEvent{type=CONNECTION_RECONNECTED data=null} >> > >> > The data monitored changed during those events, but there is not >> CHILD_UPDATED event received. >> > >> > What are the guaranties in this case? >> > Is it expected to have to fetch the data manually and do a manual diff >> when there is a reconnection? >> > >> > Thanks >> > Benjamin >> >> > >
