[
https://issues.apache.org/jira/browse/HDFS-11627?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15982294#comment-15982294
]
Mukul Kumar Singh commented on HDFS-11627:
------------------------------------------
Hi Chen,
Thanks for the review. I have addressed most of the comments
1) {code} LOG.error("read failed for BlockId: {}", blockID, ex); {code}
done, changed.
2) {code} DFS_CBLOCK_CACHE_LEVELDB_CACHE_SIZE_MB_DEFAULT) * 1024L * 1024L {code}
done, changed using OzoneConsts.
3) {code} checkExistingDirtyLog {code}
the two earlier log entires are printing trivial information and hence were
changed to debug.
the logs of most interest are of the dirty log files found, INFO is still being
used for those logs.
4) {code} testEmptyBlockBufferHandling {code}
newCache is being used, because the new instance of flusher needs the cache to
write blocks to containers. Also newCache and newFlusher along with newMetrics
help in keeping two different instances of counters.
> Block Storage: Cblock cache should register with flusher to upload blocks to
> containers
> ---------------------------------------------------------------------------------------
>
> Key: HDFS-11627
> URL: https://issues.apache.org/jira/browse/HDFS-11627
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: ozone
> Reporter: Mukul Kumar Singh
> Assignee: Mukul Kumar Singh
> Attachments: HDFS-11627-HDFS-7240.001.patch
>
>
> Cblock cache should register with flusher to upload blocks to containers.
> Currently Container Cache flusher tries to write to the container even when
> the CblockLocalCache pipelines are not registered with the flusher.
> This will result in the Container writes to fail.
> CblockLocalCache should register with the flusher before accepting any blocks
> for write
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
