[
https://issues.apache.org/jira/browse/HBASE-18090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16158178#comment-16158178
]
xinxin fan commented on HBASE-18090:
------------------------------------
Year, if the region is read-only, the writesEnabled will be false, and the
regionSequenceIdFile will not created.
{code:java}
if (this.writestate.writesEnabled) {
nextSeqid =
WALSplitter.writeRegionSequenceIdFile(this.fs.getFileSystem(), this.fs
.getRegionDir(), nextSeqid, (this.recovering ? (this.flushPerChanges
+ 10000000) : 1));
} else {
nextSeqid++;
}
{code}
And i notice the fact that a given region will be set read-only mode only when
the table is read-only or the region is not default replica. so it seem
feasible that set the HTableDescriptor read-only before the region open (in
ClientSideRegionScanner.java)：
{code:java}
public ClientSideRegionScanner(Configuration conf, FileSystem fs,
Path rootDir, HTableDescriptor htd, HRegionInfo hri, Scan scan,
ScanMetrics scanMetrics)
throws IOException {
// region is immutable, set isolation level
scan.setIsolationLevel(IsolationLevel.READ_UNCOMMITTED);
htd.setReadOnly(true);
// open region from the snapshot directory
this.region = HRegion.openHRegion(conf, fs, rootDir, hri, htd, null, null,
null);
{code}
I have tested the plan, the split tasks works well and the exception disappear.
> Improve TableSnapshotInputFormat to allow more multiple mappers per region
> --------------------------------------------------------------------------
>
> Key: HBASE-18090
> URL: https://issues.apache.org/jira/browse/HBASE-18090
> Project: HBase
> Issue Type: Improvement
> Components: mapreduce
> Affects Versions: 1.4.0
> Reporter: Mikhail Antonov
> Attachments: HBASE-18090-branch-1.3-v1.patch
>
>
> TableSnapshotInputFormat runs one map task per region in the table snapshot.
> This places unnecessary restriction that the region layout of the original
> table needs to take the processing resources available to MR job into
> consideration. Allowing to run multiple mappers per region (assuming
> reasonably even key distribution) would be useful.
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
