[
https://issues.apache.org/jira/browse/HBASE-10171?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jingcheng Du updated HBASE-10171:
---------------------------------
Description:
This umbrella JIRA includes some improvements of the coprocessor which are
listed as following.
1. Batch coprocessor: Group the executions issued to the same region server to
reduce the thread cost in the client side.
2. Coprocessor task monitor: Facilitate the monitoring of the coprocessor task
in the server side. This patch depends on the one of the batch coprocessor
since coprocessor task monitor includes the batched coprocessor tasks into the
monitoring.
3. Add a state checker for the coprocessor task: If the execution in the
coprocessor task is too long, the call in the client side will be timed out.
After that the client will retry, the timeout and retry will happen again and
again, which leads to even heavier pressure on the servers and thus to a worse
situation. This state checker could avoid the retry if the coprocessor tasks
are still in progress in the server side. This patch will be provided later and
it depends on the ones of batch coprocessor and coprocessor task monitor since
it uses the client of the coprocessor task monitor to communicate with the
coprocessor task monitor in the server side and provides the support to the
batch coprocessor.
was:
This umbrella JIRA includes some improvements of the coprocessor which are
listed as following.
1. Batch coprocessor: Group the executions issued to the same region server to
reduce the thread cost in the client side.
2. Coprocessor task monitor: Facilitate the monitoring of the coprocessor task
in the server side. This patch depends on the one of the batch coprocessor
since coprocessor task monitor includes the batched coprocessor tasks into the
monitoring.
3. Add a state checker for the coprocessor task: If the execution in the
coprocessor task is too long, the call in the client side will be timed out.
After that the client will retry, the timeout and retry will happen again and
again, which leads to even heavier pressure on the servers and thus to a worse
situation. This state checker could avoid the retry if the coprocessor tasks
are still in progress in the server side. This patch will be provided later and
it depends on the ones of batch coprocessor and coprocessor task monitor.
> Coprocessor improvement
> -----------------------
>
> Key: HBASE-10171
> URL: https://issues.apache.org/jira/browse/HBASE-10171
> Project: HBase
> Issue Type: New Feature
> Components: Coprocessors
> Affects Versions: 0.99.0
> Reporter: Jingcheng Du
> Assignee: Jingcheng Du
>
> This umbrella JIRA includes some improvements of the coprocessor which are
> listed as following.
> 1. Batch coprocessor: Group the executions issued to the same region server
> to reduce the thread cost in the client side.
> 2. Coprocessor task monitor: Facilitate the monitoring of the coprocessor
> task in the server side. This patch depends on the one of the batch
> coprocessor since coprocessor task monitor includes the batched coprocessor
> tasks into the monitoring.
> 3. Add a state checker for the coprocessor task: If the execution in the
> coprocessor task is too long, the call in the client side will be timed out.
> After that the client will retry, the timeout and retry will happen again and
> again, which leads to even heavier pressure on the servers and thus to a
> worse situation. This state checker could avoid the retry if the coprocessor
> tasks are still in progress in the server side. This patch will be provided
> later and it depends on the ones of batch coprocessor and coprocessor task
> monitor since it uses the client of the coprocessor task monitor to
> communicate with the coprocessor task monitor in the server side and provides
> the support to the batch coprocessor.
--
This message was sent by Atlassian JIRA
(v6.1.4#6159)
