I took a look last night and it sounds like the big problem is the serialization mechanism. The transportation layer has a limit of 1M in size, the getChildren method pass all the children name in that chunk, so if you have a lot of children then you are in trouble.
On Wed, Dec 5, 2012 at 7:09 AM, Bryan Thompson <[email protected]> wrote: > InterestingŠ. I had long ago requested clarification of the zookeeper > limits. This is in keeping with my expectations. Good for relatively > small numbers of children. > > For example, you would be safe trying to coordinate 1000 machines but not > safe trying to coordinate the locations of 100,000+ shards over 1000 > machines. > > Bryan > > On 12/4/12 9:42 PM, "Diego Oliveira" <[email protected]> wrote: > > >Hello folks, > > > > I got to the Curator wiki page to see something about the distributed > >queue recipe and got surprised with the big warning telling to not use > >zookeeper as a distributed queue controller, the reason page described > >some > >cases that I passed here... is this *TRUE*, if so, what option do we have. > > > >The message: > >*IMPORTANT* - We recommend that you do NOT use ZooKeeper for Queues. > >Please > >see Tech Note 4 <https://github.com/Netflix/curator/wiki/Tech-Note-4> for > >details. Links > > > >https://github.com/Netflix/curator/wiki/Distributed-Queue > >https://github.com/Netflix/curator/wiki/Tech-Note-4 > > > >-- > >Att. > >Diego de Oliveira > >System Architect > >[email protected] > >www.diegooliveira.com > >Never argue with a fool -- people might not be able to tell the difference > > -- Att. Diego de Oliveira System Architect [email protected] www.diegooliveira.com Never argue with a fool -- people might not be able to tell the difference
