[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Sam Tunnicliffe updated CASSANDRA-9669: --- Resolution: Fixed Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Status: Resolved (was: Patch Available) ok, committed Branimir's patch to 2.2 in {{66c8f2b7f79fe794cc1e0594d9add260c209a9a2}} and mine to 3.0 in {{81ffc4601952ff3a9fec8493cd27fe52544ea115}}. > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aleksey Yeschenko updated CASSANDRA-9669: - Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Status: Patch Available (was: Reopened) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-9669: Resolution: Fixed Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Status: Resolved (was: Ready to Commit) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joshua McKenzie updated CASSANDRA-9669: --- Status: Ready to Commit (was: Patch Available) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joshua McKenzie updated CASSANDRA-9669: --- Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Reviewer: Benedict > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joshua McKenzie updated CASSANDRA-9669: --- Reviewer: (was: Branimir Lambov) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joshua McKenzie updated CASSANDRA-9669: --- Assignee: Branimir Lambov (was: Benedict) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Branimir Lambov >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Jeremiah Jordan updated CASSANDRA-9669: --- Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Status: Patch Available (was: Reopened) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Sylvain Lebresne updated CASSANDRA-9669: Resolution: Fixed Assignee: Benedict Fix Version/s: (was: 3.0.x) (was: 2.2.x) (was: 3.x) 3.0.7 3.7 2.2.7 Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Status: Resolved (was: Ready to Commit) Perfect, committed, thanks. > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.7, 3.7, 3.0.7 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Branimir Lambov updated CASSANDRA-9669: --- Status: Ready to Commit (was: Patch Available) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.x, 3.0.x, 3.x > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-9669: Assignee: (was: Benedict) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.x, 3.0.x, 3.x > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] T Jake Luciani updated CASSANDRA-9669: -- Fix Version/s: (was: 3.0.1) (was: 3.1) (was: 2.2.4) 3.x 3.0.x 2.2.x > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.x, 3.0.x, 3.x > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Sylvain Lebresne updated CASSANDRA-9669: Fix Version/s: (was: 2.1.12) > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.4, 3.0.1, 3.1 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aleksey Yeschenko updated CASSANDRA-9669: - Fix Version/s: 3.0.1 > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.1.12, 2.2.4, 3.0.1, 3.1 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aleksey Yeschenko updated CASSANDRA-9669: - Fix Version/s: (was: 3.0.x) (was: 2.2.x) (was: 2.1.x) 3.1 2.1.12 2.2.4 > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.4, 2.1.12, 3.1 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Aleksey Yeschenko updated CASSANDRA-9669: - Component/s: Local Write-Read Paths > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Local Write-Read Paths >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.2.4, 2.1.12, 3.1 > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joshua McKenzie updated CASSANDRA-9669: --- Fix Version/s: (was: 3.0.0 rc2) 3.0.x 2.2.x 2.1.x > If sstable flushes complete out of order, on restart we can fail to replay > necessary commit log records > --- > > Key: CASSANDRA-9669 > URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 > Project: Cassandra > Issue Type: Bug > Components: Core >Reporter: Benedict >Assignee: Benedict >Priority: Critical > Labels: correctness > Fix For: 2.1.x, 2.2.x, 3.0.x > > > While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, > on restart we simply take the maximum replay position of any sstable on disk, > and ignore anything prior. > It is quite possible for there to be two flushes triggered for a given table, > and for the second to finish first by virtue of containing a much smaller > quantity of live data (or perhaps the disk is just under less pressure). If > we crash before the first sstable has been written, then on restart the data > it would have represented will disappear, since we will not replay the CL > records. > This looks to be a bug present since time immemorial, and also seems pretty > serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Pavel Yaskevich updated CASSANDRA-9669: --- Summary: If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records (was: Commit Log Replay is Broken) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records --- Key: CASSANDRA-9669 URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 Project: Cassandra Issue Type: Bug Components: Core Reporter: Benedict Assignee: Benedict Priority: Critical Labels: correctness Fix For: 3.x, 2.1.x, 2.2.x, 3.0.x While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, on restart we simply take the maximum replay position of any sstable on disk, and ignore anything prior. It is quite possible for there to be two flushes triggered for a given table, and for the second to finish first by virtue of containing a much smaller quantity of live data (or perhaps the disk is just under less pressure). If we crash before the first sstable has been written, then on restart the data it would have represented will disappear, since we will not replay the CL records. This looks to be a bug present since time immemorial, and also seems pretty serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Benedict updated CASSANDRA-9669: Fix Version/s: 3.0.x 2.2.x 2.1.x 3.x If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records --- Key: CASSANDRA-9669 URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 Project: Cassandra Issue Type: Bug Components: Core Reporter: Benedict Priority: Critical Labels: correctness Fix For: 3.x, 2.1.x, 2.2.x, 3.0.x While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, on restart we simply take the maximum replay position of any sstable on disk, and ignore anything prior. It is quite possible for there to be two flushes triggered for a given table, and for the second to finish first by virtue of containing a much smaller quantity of live data (or perhaps the disk is just under less pressure). If we crash before the first sstable has been written, then on restart the data it would have represented will disappear, since we will not replay the CL records. This looks to be a bug present since time immemorial, and also seems pretty serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Updated] (CASSANDRA-9669) If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records
[ https://issues.apache.org/jira/browse/CASSANDRA-9669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Michael Shuler updated CASSANDRA-9669: -- Assignee: Benedict Reproduced In: 2.1.7, 2.0.16, 1.2.19 (was: 1.2.19, 2.0.16, 2.1.7) Reviewer: Branimir Lambov If sstable flushes complete out of order, on restart we can fail to replay necessary commit log records --- Key: CASSANDRA-9669 URL: https://issues.apache.org/jira/browse/CASSANDRA-9669 Project: Cassandra Issue Type: Bug Components: Core Reporter: Benedict Assignee: Benedict Priority: Critical Labels: correctness Fix For: 3.x, 2.1.x, 2.2.x, 3.0.x While {{postFlushExecutor}} ensures it never expires CL entries out-of-order, on restart we simply take the maximum replay position of any sstable on disk, and ignore anything prior. It is quite possible for there to be two flushes triggered for a given table, and for the second to finish first by virtue of containing a much smaller quantity of live data (or perhaps the disk is just under less pressure). If we crash before the first sstable has been written, then on restart the data it would have represented will disappear, since we will not replay the CL records. This looks to be a bug present since time immemorial, and also seems pretty serious. -- This message was sent by Atlassian JIRA (v6.3.4#6332)