[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16895284#comment-16895284 ] Duo Zhang commented on HBASE-22539: --- I guess the problem is that we release the ByteBuf too earlier...

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16895272#comment-16895272 ] Wellington Chevreuil commented on HBASE-22539: -- {quote}Did the crashed region server

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16895255#comment-16895255 ] Duo Zhang commented on HBASE-22539: --- Did the crashed region server timeout on some write requests? I

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16895235#comment-16895235 ] Duo Zhang commented on HBASE-22539: --- This should be a blocker I think. So finally the problem is in

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16895128#comment-16895128 ] Wellington Chevreuil commented on HBASE-22539: -- Thanks for the comments [~openinx]

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16894914#comment-16894914 ] Anoop Sam John commented on HBASE-22539: bq. Maybe netty is reusing buffers while allocating

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16894875#comment-16894875 ] Zheng Hu commented on HBASE-22539: -- I saw the KeyValueUtil.oswrite print the row part only, and the #2

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16894709#comment-16894709 ] Wellington Chevreuil commented on HBASE-22539: -- Went through another round of log readings,

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16859944#comment-16859944 ] Anoop Sam John commented on HBASE-22539: Comparing the code in our UnsafeUtil vs DBB.get(byte[])

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856766#comment-16856766 ] Wellington Chevreuil commented on HBASE-22539: -- Thanks everyone for all the insights.

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856697#comment-16856697 ] Duo Zhang commented on HBASE-22539: --- And in fact, a DBB.get is also an Unsafe.copyMemory, you can see

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856391#comment-16856391 ] ramkrishna.s.vasudevan commented on HBASE-22539: [~wchevreuil] Have you verified that

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856277#comment-16856277 ] Zheng Hu commented on HBASE-22539: -- It's strange that ByteBufferUtils.copyFromBufferToArray would made

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16856060#comment-16856060 ] Wellington Chevreuil commented on HBASE-22539: -- {quote}Oh, good, this is a nasty bug and

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16855786#comment-16855786 ] Duo Zhang commented on HBASE-22539: --- And in general, both SimpleRpcServer and NettyRpcServer will use

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16855736#comment-16855736 ] Duo Zhang commented on HBASE-22539: --- And could you reproduce the problem? Could you please try moving

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16855721#comment-16855721 ] Duo Zhang commented on HBASE-22539: --- Oh, good, this is a nasty bug and seems we are about to reach the

[jira] [Commented] (HBASE-22539) Potential WAL corruption due to Unsafe.copyMemory usage when DBB are in place

[ https://issues.apache.org/jira/browse/HBASE-22539?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16855688#comment-16855688 ] Wellington Chevreuil commented on HBASE-22539: -- Symptoms: Upon an RS crash that required