[
https://issues.apache.org/jira/browse/ZOOKEEPER-1962?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15080855#comment-15080855
]
Enis Soztutar commented on ZOOKEEPER-1962:
------------------------------------------
I think the {{jute.maxBuffer}} is not relevant since the recursive {{ls}}
accumulation is performed at the client side. What [~eribeiro] was referring to
is a design where ls result accumulation is performed at the server side, and
then the returned result set is returned as a whole in a single RPC.
This client side design is not a snapshot view of the whole tree, but I think
as long as the semantics are clear, it should be fine. It is a cli tool, mainly
for debugging.
One minor improvement might be that instead of accumulating the full set of
children in a {{List<String>}}, we should do incremental printing of the
results without the full list. I had to debug a multi-GB zk snapshot with 5M
znodes recently, and had trouble with this kind of
accumulate-everything-then-print strategy since it takes way too long.
> Add a CLI command to recursively list a znode and children
> ----------------------------------------------------------
>
> Key: ZOOKEEPER-1962
> URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1962
> Project: ZooKeeper
> Issue Type: New Feature
> Components: java client
> Affects Versions: 3.4.6
> Reporter: Gautam Gopalakrishnan
> Assignee: Gautam Gopalakrishnan
> Priority: Minor
> Fix For: 3.5.2, 3.6.0
>
> Attachments: ZOOKEEPER-1962.diff, ZOOKEEPER-1962_v2.patch,
> ZOOKEEPER-1962_v3.patch, ZOOKEEPER-1962_v4.patch
>
> Original Estimate: 24h
> Remaining Estimate: 24h
>
> When troubleshooting applications where znodes can be multiple levels deep
> (eg. HBase replication), it is handy to see all child znodes recursively
> rather than run an ls for each node manually.
> So I propose adding an option to the "ls" command (-r) which will list all
> child nodes under a given znode.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
