[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-02-03 Thread Bahram Chehrazy (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16759593#comment-16759593 ] Bahram Chehrazy commented on HBASE-21601: - Yeah, if the WAL is already corrupted that could be a

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-02-01 Thread Sergey Shelukhin (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16758575#comment-16758575 ] Sergey Shelukhin commented on HBASE-21601: -- The original problem is with an existing WAL before

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-02-01 Thread Bahram Chehrazy (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16758109#comment-16758109 ] Bahram Chehrazy commented on HBASE-21601: - EntryBuffers.getChunkToWrite is not thread safe and

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-02-01 Thread Duo Zhang (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16758111#comment-16758111 ] Duo Zhang commented on HBASE-21601: --- {quote} EntryBuffers.getChunkToWrite is not thread safe and is

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-01-31 Thread Duo Zhang (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757958#comment-16757958 ] Duo Zhang commented on HBASE-21601: --- And the problem is that, the WALKey is correct, but the WALEntry

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-01-31 Thread Duo Zhang (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757957#comment-16757957 ] Duo Zhang commented on HBASE-21601: --- So the correct and corrupt entries are in the same file?

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-01-31 Thread Sergey Shelukhin (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757894#comment-16757894 ] Sergey Shelukhin commented on HBASE-21601: -- The file was healthy on HDFS at the time that I

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record

2019-01-31 Thread Duo Zhang (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757893#comment-16757893 ] Duo Zhang commented on HBASE-21601: --- And could you please check whether the different blocks are in

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record - step 1, skip such records

2019-01-31 Thread Sergey Shelukhin (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757886#comment-16757886 ] Sergey Shelukhin commented on HBASE-21601: -- Given the nature of the issue it's not possible to

[jira] [Commented] (HBASE-21601) HBase (asyncwal?) can write a corrupted WAL record - step 1, skip such records

2019-01-31 Thread Duo Zhang (JIRA)
[ https://issues.apache.org/jira/browse/HBASE-21601?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16757881#comment-16757881 ] Duo Zhang commented on HBASE-21601: --- Is it possible to reproduce the problem in production? If so you