Home | About | Sematext search-lucene.com search-hadoop.com
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB
 Search Hadoop and all its subprojects:

Switch to Plain View
Zookeeper >> mail # dev >> question about getChildren and queue recipe


+
Alexander Shraer 2012-09-04, 01:10
+
Flavio Junqueira 2012-09-04, 22:11
+
Alexander Shraer 2012-09-04, 23:34
+
Flavio Junqueira 2012-09-05, 11:24
Copy link to this message
-
Re: question about getChildren and queue recipe
Hi Flavio,

Yes, an ArrayList for example, sorted by creation order. Even
currently although DataNode stores it as a Set, when you retrieve the
list of children DataTree will convert the set to an ArrayList and
return a list. Obviously exists/create will have to traverse the list
to check if it already contains something, so it may be a bit slower,
but having a list may have lots of advantages (such as saving the sort
from clients in many use cases and potentially allowing us to add a
parameter to getChildren to retrieve only the first k znodes, which is
what Jordan meant I think). I'm not sure whether we'll be able to
store in sorted form or sort every time we load the database on
recovery, and probably there are other implications, but it seems
worthwhile to consider.

Alex

On Wed, Sep 5, 2012 at 4:24 AM, Flavio Junqueira <[EMAIL PROTECTED]> wrote:
> Hi Alex, I believe we currently use a set to hold the children of a znode. Are you proposing to use a different data structure?
>
> -Flavio
>
> On Sep 5, 2012, at 1:34 AM, Alexander Shraer wrote:
>
>> Thanks Flavio. But what about just storing the list of children on the
>> server in the order of their creation ? Wouldn't this be consistent
>> with the sequential id and also cheap ? or am I missing something
>>
>> Thanks,
>> Alex
>>
>> On Tue, Sep 4, 2012 at 3:11 PM, Flavio Junqueira <[EMAIL PROTECTED]> wrote:
>>>
>>> On Sep 4, 2012, at 3:10 AM, Alexander Shraer wrote:
>>>
>>>> Hi,
>>>>
>>>> I have 2 questions:
>>>> - why was it decided not to guarantee any order on the list returned from
>>>> getChildren, given that many use-cases require looking on the child with
>>>> the smallest id ?
>>>> Why not just keep this list in a sorted form on the server ?
>>>
>>>
>>> There is some discussion about this here:
>>>
>>> https://issues.apache.org/jira/browse/ZOOKEEPER-44
>>>
>>> -Flavio
>>>
>
+
Flavio Junqueira 2012-09-06, 11:18
+
charles charles 2012-09-05, 00:32
+
Jordan Zimmerman 2012-09-05, 00:33
NEW: Monitor These Apps!
elasticsearch, apache solr, apache hbase, hadoop, redis, casssandra, amazon cloudwatch, mysql, memcached, apache kafka, apache zookeeper, apache storm, ubuntu, centOS, red hat, debian, puppet labs, java, senseiDB