[
https://issues.apache.org/jira/browse/HBASE-10932?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13981441#comment-13981441
]
Yu Li commented on HBASE-10932:
-------------------------------
Hi [~jdcryans],
If we follow this logic, do you mean the "-m" option of DistCp also useless?
IMHO, the configuration of job scheduler in JT/Yarn is the server-side
configuration, while the "-m" option is the client-side configuration, and both
are necessary.
Back to the scheduler discussion, I believe job scheduler could only limit the
max resource one user could use, and it depends on the user to decide how he
uses the resource assigned to him. Like in the example you gave, what if the
"slow" pool have 4 slots while only one user submit a rowcounter and he prefers
only 2 maps running in parallel? I'm afraid asking the cluster operator to
create another "slow" pool with only 2 slots is not a good solution.
In a common hbase ETL application, user would need to first do distcp, then
bulkload, then rowcounter to check data integrity, and he would prefer distcp
to run as fast as possible w/ low scan workload during rowcounter. In this
case, he would need to submit the distcp job to the "fast" queue while the
rowcounter job to the "slow" queue? And he also needs to get access to both
queues...
Anyway, this is a real requirement from user in our product env, and I'm just
trying to contribute this to community in case this can help other users. But
if you still think it useless, just go ahead and close it, you're the boss
after all. :-)
And no matter what decision made, thanks for your time on reviewing this JIRA
and discussion.
> Improve RowCounter to allow mapper number set/control
> -----------------------------------------------------
>
> Key: HBASE-10932
> URL: https://issues.apache.org/jira/browse/HBASE-10932
> Project: HBase
> Issue Type: Improvement
> Components: mapreduce
> Reporter: Yu Li
> Assignee: Yu Li
> Priority: Minor
> Attachments: HBASE-10932_v1.patch, HBASE-10932_v2.patch
>
>
> The typical use case of RowCounter is to do some kind of data integrity
> checking, like after exporting some data from RDBMS to HBase, or from one
> HBase cluster to another, making sure the row(record) number matches. Such
> check commonly won't require much on response time.
> Meanwhile, based on current impl, RowCounter will launch one mapper per
> region, and each mapper will send one scan request. Assuming the table is
> kind of big like having tens of regions, and the cpu core number of the whole
> MR cluster is also enough, the parallel scan requests sent by mapper would be
> a real burden for the HBase cluster.
> So in this JIRA, we're proposing to make rowcounter support an additional
> option "--maps" to specify mapper number, and make each mapper able to scan
> more than one region of the target table.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
