[jira] [Updated] (PHOENIX-5112) Simplify QueryPlan selection in Phoenix.
[
https://issues.apache.org/jira/browse/PHOENIX-5112?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Lars Hofhansl updated PHOENIX-5112:
---
Attachment: 5112-master.txt
> Simplify QueryPlan selection in Phoenix.
>
>
> Key: PHOENIX-5112
> URL: https://issues.apache.org/jira/browse/PHOENIX-5112
> Project: Phoenix
> Issue Type: Improvement
>Reporter: Lars Hofhansl
>Priority: Major
> Attachments: 5112-master.txt
>
>
> Brought to light in part in PHOENIX-5109, plan selection in Phoenix is too
> complicated with its logic spread over multiple areas. My recent changes let
> most index through the initial filters and next step is to put all the logic
> in {{QueryOptimizer.orderPlansBestToWorst}}.
> One exception is plans with global indexes for queries with uncovered
> queries, those should still be handled in {{addPlan}}.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Updated] (PHOENIX-5080) Index becomes Active during Partial Index Rebuilder if Index Failure happens
[
https://issues.apache.org/jira/browse/PHOENIX-5080?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Monani Mihir updated PHOENIX-5080:
--
Attachment: PHOENIX-5080.01.patch
> Index becomes Active during Partial Index Rebuilder if Index Failure happens
>
>
> Key: PHOENIX-5080
> URL: https://issues.apache.org/jira/browse/PHOENIX-5080
> Project: Phoenix
> Issue Type: Bug
>Affects Versions: 4.14.1
>Reporter: Monani Mihir
>Assignee: Monani Mihir
>Priority: Blocker
> Attachments: PHOENIX-5080-4.x-HBase-1.3.01.patch,
> PHOENIX-5080-4.x-HBase-1.3.02.patch, PHOENIX-5080-4.x-HBase-1.3.02.patch,
> PHOENIX-5080-4.x-HBase-1.3.03.patch, PHOENIX-5080-4.x-HBase-1.3.04.patch,
> PHOENIX-5080-4.x-HBase-1.3.05.patch, PHOENIX-5080.01.patch,
> PHOENIX-5080.01.patch
>
>
> After PHOENIX-4130 and PHOENIX-4600 , If there is Index failure during
> Partial Index Rebuild, Rebuilder will try again to write Index updates. If it
> succeeds then it will transition Index from INACTIVE to ACTIVE, even before
> Rebuilder finishes.
> Here is where it goes wrong, I think :-
> {code:java}
> PhoenixIndexFailurePolicy.java :-
> public static void doBatchWithRetries(MutateCommand mutateCommand,
> IndexWriteException iwe, PhoenixConnection connection,
> ReadOnlyProps config) throws IOException {
>
> while (canRetryMore(numRetry++, maxTries, canRetryUntil)) {
> ...
> handleIndexWriteSuccessFromClient(iwe, connection);
> ...
> }
> }
>
> private static void handleIndexWriteSuccessFromClient(IndexWriteException
> indexWriteException, PhoenixConnection conn) {
> handleExceptionFromClient(indexWriteException, conn,
> PIndexState.ACTIVE);
> }
> {code}
>
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Updated] (PHOENIX-5018) Index mutations created by UPSERT SELECT will have wrong timestamps
[ https://issues.apache.org/jira/browse/PHOENIX-5018?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kadir OZDEMIR updated PHOENIX-5018: --- Description: When doing a full rebuild (or initial async build) of a local or global index using IndexTool and PhoenixIndexImportDirectMapper, or doing a synchronous initial build of a global index using the index create DDL, we generate the index mutations by using an UPSERT SELECT query from the base table to the index. The timestamps of the mutations use the default HBase behavior, which is to take the current wall clock. However, the timestamp of an index KeyValue should use the timestamp of the initial KeyValue in the base table. Having base table and index timestamps out of sync can cause all sorts of weird side effects, such as if the base table has data with an expired TTL that isn't expired in the index yet. Also inserting old mutations with new timestamps may overwrite the data that has been newly overwritten by the regular data path during index build, which would lead to data loss and inconsistency issues. was: When doing a full rebuild (or initial async build) on an index using the IndexTool and PhoenixIndexImportDirectMapper, we generate the index mutations by creating an UPSERT SELECT query from the base table to the index, then taking the Mutations from it and inserting it directly into the index via an HBase HTable. The timestamps of the Mutations use the default HBase behavior, which is to take the current wall clock. However, the timestamp of an index KeyValue should use the timestamp of the initial KeyValue in the base table. Having base table and index timestamps out of sync can cause all sorts of weird side effects, such as if the base table has data with an expired TTL that isn't expired in the index yet. > Index mutations created by UPSERT SELECT will have wrong timestamps > --- > > Key: PHOENIX-5018 > URL: https://issues.apache.org/jira/browse/PHOENIX-5018 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.14.0, 5.0.0 >Reporter: Geoffrey Jacoby >Assignee: Kadir OZDEMIR >Priority: Major > > When doing a full rebuild (or initial async build) of a local or global index > using IndexTool and PhoenixIndexImportDirectMapper, or doing a synchronous > initial build of a global index using the index create DDL, we generate the > index mutations by using an UPSERT SELECT query from the base table to the > index. > The timestamps of the mutations use the default HBase behavior, which is to > take the current wall clock. However, the timestamp of an index KeyValue > should use the timestamp of the initial KeyValue in the base table. > Having base table and index timestamps out of sync can cause all sorts of > weird side effects, such as if the base table has data with an expired TTL > that isn't expired in the index yet. Also inserting old mutations with new > timestamps may overwrite the data that has been newly overwritten by the > regular data path during index build, which would lead to data loss and > inconsistency issues. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5018) Index mutations created by UPSERT SELECT will have wrong timestamps
[ https://issues.apache.org/jira/browse/PHOENIX-5018?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kadir OZDEMIR updated PHOENIX-5018: --- Summary: Index mutations created by UPSERT SELECT will have wrong timestamps (was: Index mutations created by a full build will have wrong timestamps) > Index mutations created by UPSERT SELECT will have wrong timestamps > --- > > Key: PHOENIX-5018 > URL: https://issues.apache.org/jira/browse/PHOENIX-5018 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.14.0, 5.0.0 >Reporter: Geoffrey Jacoby >Assignee: Kadir OZDEMIR >Priority: Major > > When doing a full rebuild (or initial async build) on an index using the > IndexTool and PhoenixIndexImportDirectMapper, we generate the index mutations > by creating an UPSERT SELECT query from the base table to the index, then > taking the Mutations from it and inserting it directly into the index via an > HBase HTable. > The timestamps of the Mutations use the default HBase behavior, which is to > take the current wall clock. However, the timestamp of an index KeyValue > should use the timestamp of the initial KeyValue in the base table. > Having base table and index timestamps out of sync can cause all sorts of > weird side effects, such as if the base table has data with an expired TTL > that isn't expired in the index yet. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5018) Index mutations created by a full build will have wrong timestamps
[ https://issues.apache.org/jira/browse/PHOENIX-5018?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kadir OZDEMIR updated PHOENIX-5018: --- Summary: Index mutations created by a full build will have wrong timestamps (was: Index mutations created by IndexTool will have wrong timestamps) > Index mutations created by a full build will have wrong timestamps > -- > > Key: PHOENIX-5018 > URL: https://issues.apache.org/jira/browse/PHOENIX-5018 > Project: Phoenix > Issue Type: Bug >Affects Versions: 4.14.0, 5.0.0 >Reporter: Geoffrey Jacoby >Assignee: Kadir OZDEMIR >Priority: Major > > When doing a full rebuild (or initial async build) on an index using the > IndexTool and PhoenixIndexImportDirectMapper, we generate the index mutations > by creating an UPSERT SELECT query from the base table to the index, then > taking the Mutations from it and inserting it directly into the index via an > HBase HTable. > The timestamps of the Mutations use the default HBase behavior, which is to > take the current wall clock. However, the timestamp of an index KeyValue > should use the timestamp of the initial KeyValue in the base table. > Having base table and index timestamps out of sync can cause all sorts of > weird side effects, such as if the base table has data with an expired TTL > that isn't expired in the index yet. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5080) Index becomes Active during Partial Index Rebuilder if Index Failure happens
[
https://issues.apache.org/jira/browse/PHOENIX-5080?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Monani Mihir updated PHOENIX-5080:
--
Attachment: PHOENIX-5080.01.patch
> Index becomes Active during Partial Index Rebuilder if Index Failure happens
>
>
> Key: PHOENIX-5080
> URL: https://issues.apache.org/jira/browse/PHOENIX-5080
> Project: Phoenix
> Issue Type: Bug
>Affects Versions: 4.14.1
>Reporter: Monani Mihir
>Assignee: Monani Mihir
>Priority: Blocker
> Attachments: PHOENIX-5080-4.x-HBase-1.3.01.patch,
> PHOENIX-5080-4.x-HBase-1.3.02.patch, PHOENIX-5080-4.x-HBase-1.3.02.patch,
> PHOENIX-5080-4.x-HBase-1.3.03.patch, PHOENIX-5080-4.x-HBase-1.3.04.patch,
> PHOENIX-5080-4.x-HBase-1.3.05.patch, PHOENIX-5080.01.patch
>
>
> After PHOENIX-4130 and PHOENIX-4600 , If there is Index failure during
> Partial Index Rebuild, Rebuilder will try again to write Index updates. If it
> succeeds then it will transition Index from INACTIVE to ACTIVE, even before
> Rebuilder finishes.
> Here is where it goes wrong, I think :-
> {code:java}
> PhoenixIndexFailurePolicy.java :-
> public static void doBatchWithRetries(MutateCommand mutateCommand,
> IndexWriteException iwe, PhoenixConnection connection,
> ReadOnlyProps config) throws IOException {
>
> while (canRetryMore(numRetry++, maxTries, canRetryUntil)) {
> ...
> handleIndexWriteSuccessFromClient(iwe, connection);
> ...
> }
> }
>
> private static void handleIndexWriteSuccessFromClient(IndexWriteException
> indexWriteException, PhoenixConnection conn) {
> handleExceptionFromClient(indexWriteException, conn,
> PIndexState.ACTIVE);
> }
> {code}
>
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Updated] (PHOENIX-5080) Index becomes Active during Partial Index Rebuilder if Index Failure happens
[
https://issues.apache.org/jira/browse/PHOENIX-5080?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Monani Mihir updated PHOENIX-5080:
--
Attachment: PHOENIX-5080-4.x-HBase-1.3.05.patch
> Index becomes Active during Partial Index Rebuilder if Index Failure happens
>
>
> Key: PHOENIX-5080
> URL: https://issues.apache.org/jira/browse/PHOENIX-5080
> Project: Phoenix
> Issue Type: Bug
>Affects Versions: 4.14.1
>Reporter: Monani Mihir
>Assignee: Monani Mihir
>Priority: Blocker
> Attachments: PHOENIX-5080-4.x-HBase-1.3.01.patch,
> PHOENIX-5080-4.x-HBase-1.3.02.patch, PHOENIX-5080-4.x-HBase-1.3.02.patch,
> PHOENIX-5080-4.x-HBase-1.3.03.patch, PHOENIX-5080-4.x-HBase-1.3.04.patch,
> PHOENIX-5080-4.x-HBase-1.3.05.patch
>
>
> After PHOENIX-4130 and PHOENIX-4600 , If there is Index failure during
> Partial Index Rebuild, Rebuilder will try again to write Index updates. If it
> succeeds then it will transition Index from INACTIVE to ACTIVE, even before
> Rebuilder finishes.
> Here is where it goes wrong, I think :-
> {code:java}
> PhoenixIndexFailurePolicy.java :-
> public static void doBatchWithRetries(MutateCommand mutateCommand,
> IndexWriteException iwe, PhoenixConnection connection,
> ReadOnlyProps config) throws IOException {
>
> while (canRetryMore(numRetry++, maxTries, canRetryUntil)) {
> ...
> handleIndexWriteSuccessFromClient(iwe, connection);
> ...
> }
> }
>
> private static void handleIndexWriteSuccessFromClient(IndexWriteException
> indexWriteException, PhoenixConnection conn) {
> handleExceptionFromClient(indexWriteException, conn,
> PIndexState.ACTIVE);
> }
> {code}
>
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
[jira] [Updated] (PHOENIX-5094) Index can transition from INACTIVE to ACTIVE via Phoenix Client
[ https://issues.apache.org/jira/browse/PHOENIX-5094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kiran Kumar Maturi updated PHOENIX-5094: Attachment: PHOENIX-5094-4.14-HBase-1.3.04.patch > Index can transition from INACTIVE to ACTIVE via Phoenix Client > --- > > Key: PHOENIX-5094 > URL: https://issues.apache.org/jira/browse/PHOENIX-5094 > Project: Phoenix > Issue Type: Bug >Affects Versions: 5.0.0, 4.14.1 >Reporter: Monani Mihir >Assignee: Kiran Kumar Maturi >Priority: Major > Attachments: PHOENIX-5094-4.14-HBase-1.3.01.patch, > PHOENIX-5094-4.14-HBase-1.3.02.patch, PHOENIX-5094-4.14-HBase-1.3.03.patch, > PHOENIX-5094-4.14-HBase-1.3.04.patch, PHOENIX-5094-master.01.patch, > PHOENIX-5094-master.02.patch > > > Suppose Index is in INACTIVE state and Client load is running continuously. > With INACTIVE State, client will keep maintaining index. > Before Rebuilder could run and bring index back in sync with data table, If > some mutation for Index fails from client side, then client will transition > Index state (From INACTIVE--> PENDING_DISABLE). > If client succeeds in writing mutation in subsequent retries, it will > transition Index state again ( From PENDING_DISABLE --> ACTIVE) . > Above scenario will leave some part of Index out of sync with data table. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5094) Index can transition from INACTIVE to ACTIVE via Phoenix Client
[ https://issues.apache.org/jira/browse/PHOENIX-5094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kiran Kumar Maturi updated PHOENIX-5094: Attachment: (was: PHOENIX-5094-4.14-HBase-1.3.03.patch) > Index can transition from INACTIVE to ACTIVE via Phoenix Client > --- > > Key: PHOENIX-5094 > URL: https://issues.apache.org/jira/browse/PHOENIX-5094 > Project: Phoenix > Issue Type: Bug >Affects Versions: 5.0.0, 4.14.1 >Reporter: Monani Mihir >Assignee: Kiran Kumar Maturi >Priority: Major > Attachments: PHOENIX-5094-4.14-HBase-1.3.01.patch, > PHOENIX-5094-4.14-HBase-1.3.02.patch, PHOENIX-5094-4.14-HBase-1.3.03.patch, > PHOENIX-5094-4.14-HBase-1.3.04.patch, PHOENIX-5094-master.01.patch, > PHOENIX-5094-master.02.patch > > > Suppose Index is in INACTIVE state and Client load is running continuously. > With INACTIVE State, client will keep maintaining index. > Before Rebuilder could run and bring index back in sync with data table, If > some mutation for Index fails from client side, then client will transition > Index state (From INACTIVE--> PENDING_DISABLE). > If client succeeds in writing mutation in subsequent retries, it will > transition Index state again ( From PENDING_DISABLE --> ACTIVE) . > Above scenario will leave some part of Index out of sync with data table. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5094) Index can transition from INACTIVE to ACTIVE via Phoenix Client
[ https://issues.apache.org/jira/browse/PHOENIX-5094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kiran Kumar Maturi updated PHOENIX-5094: Attachment: PHOENIX-5094-4.14-HBase-1.3.03.patch > Index can transition from INACTIVE to ACTIVE via Phoenix Client > --- > > Key: PHOENIX-5094 > URL: https://issues.apache.org/jira/browse/PHOENIX-5094 > Project: Phoenix > Issue Type: Bug >Affects Versions: 5.0.0, 4.14.1 >Reporter: Monani Mihir >Assignee: Kiran Kumar Maturi >Priority: Major > Attachments: PHOENIX-5094-4.14-HBase-1.3.01.patch, > PHOENIX-5094-4.14-HBase-1.3.02.patch, PHOENIX-5094-4.14-HBase-1.3.03.patch, > PHOENIX-5094-4.14-HBase-1.3.03.patch, PHOENIX-5094-master.01.patch, > PHOENIX-5094-master.02.patch > > > Suppose Index is in INACTIVE state and Client load is running continuously. > With INACTIVE State, client will keep maintaining index. > Before Rebuilder could run and bring index back in sync with data table, If > some mutation for Index fails from client side, then client will transition > Index state (From INACTIVE--> PENDING_DISABLE). > If client succeeds in writing mutation in subsequent retries, it will > transition Index state again ( From PENDING_DISABLE --> ACTIVE) . > Above scenario will leave some part of Index out of sync with data table. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5094) Index can transition from INACTIVE to ACTIVE via Phoenix Client
[ https://issues.apache.org/jira/browse/PHOENIX-5094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kiran Kumar Maturi updated PHOENIX-5094: Attachment: PHOENIX-5094-master.02.patch > Index can transition from INACTIVE to ACTIVE via Phoenix Client > --- > > Key: PHOENIX-5094 > URL: https://issues.apache.org/jira/browse/PHOENIX-5094 > Project: Phoenix > Issue Type: Bug >Affects Versions: 5.0.0, 4.14.1 >Reporter: Monani Mihir >Assignee: Kiran Kumar Maturi >Priority: Major > Attachments: PHOENIX-5094-4.14-HBase-1.3.01.patch, > PHOENIX-5094-4.14-HBase-1.3.02.patch, PHOENIX-5094-4.14-HBase-1.3.03.patch, > PHOENIX-5094-4.14-HBase-1.3.03.patch, PHOENIX-5094-master.01.patch, > PHOENIX-5094-master.02.patch > > > Suppose Index is in INACTIVE state and Client load is running continuously. > With INACTIVE State, client will keep maintaining index. > Before Rebuilder could run and bring index back in sync with data table, If > some mutation for Index fails from client side, then client will transition > Index state (From INACTIVE--> PENDING_DISABLE). > If client succeeds in writing mutation in subsequent retries, it will > transition Index state again ( From PENDING_DISABLE --> ACTIVE) . > Above scenario will leave some part of Index out of sync with data table. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-5094) Index can transition from INACTIVE to ACTIVE via Phoenix Client
[ https://issues.apache.org/jira/browse/PHOENIX-5094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Kiran Kumar Maturi updated PHOENIX-5094: Attachment: PHOENIX-5094-4.14-HBase-1.3.03.patch > Index can transition from INACTIVE to ACTIVE via Phoenix Client > --- > > Key: PHOENIX-5094 > URL: https://issues.apache.org/jira/browse/PHOENIX-5094 > Project: Phoenix > Issue Type: Bug >Affects Versions: 5.0.0, 4.14.1 >Reporter: Monani Mihir >Assignee: Kiran Kumar Maturi >Priority: Major > Attachments: PHOENIX-5094-4.14-HBase-1.3.01.patch, > PHOENIX-5094-4.14-HBase-1.3.02.patch, PHOENIX-5094-4.14-HBase-1.3.03.patch, > PHOENIX-5094-master.01.patch > > > Suppose Index is in INACTIVE state and Client load is running continuously. > With INACTIVE State, client will keep maintaining index. > Before Rebuilder could run and bring index back in sync with data table, If > some mutation for Index fails from client side, then client will transition > Index state (From INACTIVE--> PENDING_DISABLE). > If client succeeds in writing mutation in subsequent retries, it will > transition Index state again ( From PENDING_DISABLE --> ACTIVE) . > Above scenario will leave some part of Index out of sync with data table. -- This message was sent by Atlassian JIRA (v7.6.3#76005)
[jira] [Updated] (PHOENIX-4993) Data table region should not close RS level shared/cached connections like IndexWriter, RecoveryIndexWriter
[
https://issues.apache.org/jira/browse/PHOENIX-4993?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Kiran Kumar Maturi updated PHOENIX-4993:
Attachment: PHOENIX-4993-master.02.patch
> Data table region should not close RS level shared/cached connections like
> IndexWriter, RecoveryIndexWriter
> ---
>
> Key: PHOENIX-4993
> URL: https://issues.apache.org/jira/browse/PHOENIX-4993
> Project: Phoenix
> Issue Type: Bug
>Affects Versions: 4.14.0
>Reporter: Kiran Kumar Maturi
>Assignee: Kiran Kumar Maturi
>Priority: Major
> Fix For: 4.15.0, 5.1.0
>
> Attachments: PHOENIX-4993-4.x-HBase-1.3.01.patch,
> PHOENIX-4993-4.x-HBase-1.3.02.patch, PHOENIX-4993-4.x-HBase-1.3.03.patch,
> PHOENIX-4993-4.x-HBase-1.3.04.patch, PHOENIX-4993-4.x-HBase-1.3.05.patch,
> PHOENIX-4993-master.01.patch, PHOENIX-4993-master.02.patch
>
>
> Issue is related to Region Server being killed when one region is closing and
> another region is trying to write index updates.
> When the data table region closes it will close region server level
> cached/shared connections and it could interrupt other region
> index/index-state update.
> -- Region1: Closing
> {code:java}
> TrackingParallellWriterIndexCommitter#stop() {
> this.retryingFactory.shutdown();
> this.noRetriesFactory.shutdown();
> }{code}
> closes the cached connections calling
> CoprocessorHConnectionTableFactory#shutdown() in ServerUtil.java
>
> --Region2: Writing index updates
> Index updates fail as connections are closed, which leads to
> RejectedExecutionException/Connection being null. This triggers
> PhoenixIndexFailurePolicy#handleFailureWithExceptions that tries to get the
> the syscat table using the cached connections. Here it will not be able to
> reach to SYSCAT , so we will trigger KillServreFailurePolicy.
> CoprocessorHConnectionTableFactory#getTable()
>
>
> {code:java}
> if (connection == null || connection.isClosed()) {
> throw new IllegalArgumentException("Connection is null or closed.");
> }{code}
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
