[
https://issues.apache.org/jira/browse/HBASE-17372?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15780055#comment-15780055
]
Duo Zhang commented on HBASE-17372:
-----------------------------------
For now the read timeout is for exists, scan and get, and write timeout is for
all other operations. For batch, if a group only have Get then we use read
timeout, if no Get then we use write timeout, if both then we will use the
larger one.
Agree that we'd better use a single rpc timeout as the user may not set the
correct timeout, for example, set read timeout but calling checkAndPut. Then
how do we deal with the default rpc timeout of batch? A new timeout config for
batch? Then should multiGet, multiPut use the batch timeout?
Yeah maybe we do not need to introduce so many rpc timeout configs as it could
be changed per call now...
> Make AsyncTable thread safe
> ---------------------------
>
> Key: HBASE-17372
> URL: https://issues.apache.org/jira/browse/HBASE-17372
> Project: HBase
> Issue Type: Sub-task
> Components: asyncclient, Client
> Affects Versions: 2.0.0
> Reporter: Duo Zhang
> Assignee: Duo Zhang
> Fix For: 2.0.0
>
> Attachments: HBASE-17372.patch
>
>
> The most methods are already thread safe. The problem is that we have some
> methods that used to set timeout, we need to remove these methods and add a
> parameter for each call to specific timeout settings.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
