[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16543312#comment-16543312 ] James Taylor commented on OMID-103: --- You could pretty easily convert TransactionIT.testFailureToRollbackAfterDelete() to a standalone HBase test to create a unit test for OMID-102, [~ohads]. > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Assignee: Ohad Shacham >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16540192#comment-16540192 ] Ohad Shacham commented on OMID-103: --- Not sure how the filter semantics matches the semantics of select *. Are you looking at the uses of the results? > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16540177#comment-16540177 ] Ohad Shacham commented on OMID-103: --- [~jamestaylor], it seems that the problem caused by the FirstKeyOnlyFilter that the scan includes. It limits the number of cells returns by the scan and Omid can't identify the snapshot this way. It will be solved by OMID-102. What is the reason for including FirstKeyOnlyFilter for select * from table? > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16539703#comment-16539703 ] James Taylor commented on OMID-103: --- One thing I left out: the subsequent attempt to abort the transaction fails in this test. This would typically undo all the writes that were done through HBase by issuing point deletes. In this case, those deletes don't occur and subsequent reads of the table do not do the correct filtering. I think that's the crux of the issue, [~ohads]. > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16538145#comment-16538145 ] James Taylor commented on OMID-103: --- Sorry for the confusion, [~ohads]. The scenario is: * Insert a new row * Commit the transaction * Delete the row (forcing delete to be written to HBase) * Insert a new row * Attempt to commit the transaction (which will fail) * Query the table and the delete will still be in effect even though the commit failed Maybe it is related to Omid not having a batch operation for its HTable implementation, since in the above case the Delete and Put would come through separate HBase calls? > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[
https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16538104#comment-16538104
]
Ohad Shacham commented on OMID-103:
---
[~jamestaylor], this is not what I understand from the code.
I thought that the problem is that the commit fails and still the deletion
marker is visible.
It assumes that the commit fails:
try {
conn.commit();
fail();
} catch (SQLException e) {
}
actually, it doesn't even call Omid's commit.
What am I missing?
> DeleteFamily marker still visible after rollback fails
> --
>
> Key: OMID-103
> URL: https://issues.apache.org/jira/browse/OMID-103
> Project: Apache Omid
> Issue Type: Bug
>Reporter: James Taylor
>Priority: Major
>
> If a DeleteFamily marker is written to HBase and the subsequent abort of the
> transaction fails, the delete marker is still visible and mistakenly
> continues to take effect. For a repro, see
> TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but
> fails for Omid.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Commented] (OMID-103) DeleteFamily marker still visible after rollback fails
[ https://issues.apache.org/jira/browse/OMID-103?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16537717#comment-16537717 ] James Taylor commented on OMID-103: --- FYI, [~ohads] - the specific scenario in this case is: * HBase write of family delete marker occurs successfully * Commit was successful * Write of shadow cells failed * Query/scan over rows doesn't properly filter out all deleted cells for the family delete marker > DeleteFamily marker still visible after rollback fails > -- > > Key: OMID-103 > URL: https://issues.apache.org/jira/browse/OMID-103 > Project: Apache Omid > Issue Type: Bug >Reporter: James Taylor >Priority: Major > > If a DeleteFamily marker is written to HBase and the subsequent abort of the > transaction fails, the delete marker is still visible and mistakenly > continues to take effect. For a repro, see > TransactionIT.testFailureToRollbackAfterDelete() which passes for Tephra but > fails for Omid. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
