[
https://issues.apache.org/jira/browse/TAJO-1412?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14498250#comment-14498250
]
ASF GitHub Bot commented on TAJO-1412:
--------------------------------------
Github user jihoonson commented on the pull request:
https://github.com/apache/tajo/pull/432#issuecomment-93780346
Hi @navis. I've reviewed this patch carefully.
Sorry to say, but I think that this is not appropriate for our architecture.
When a client calls ```pollQueryStatus()```, the process who waits for the
query state to be changed is ```TajoMaster```'s rpc server. As you know, we
will support multi-tenancy soon. In multi-tenant environment, master's rpc
server will be getting slower as more clients submit queries and wait for their
results. IMHO, polling in only client will be better than blocking master.
What do you think?
> Add polling API in TajoMasterClientProtocol to wait result effectively
> ----------------------------------------------------------------------
>
> Key: TAJO-1412
> URL: https://issues.apache.org/jira/browse/TAJO-1412
> Project: Tajo
> Issue Type: Improvement
> Components: client
> Reporter: Navis
> Assignee: Navis
> Priority: Trivial
>
> Currently, client polls master node to acquire result frequently (every 500
> msec), which seemed not small enough for light queries and not big enough for
> heavy queries.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
