[jira] [Commented] (FLINK-8322) support getting number of existing timers in TimerService
[ https://issues.apache.org/jira/browse/FLINK-8322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324977#comment-16324977 ] ASF GitHub Bot commented on FLINK-8322: --- Github user bowenli86 closed the pull request at: https://github.com/apache/flink/pull/5236 > support getting number of existing timers in TimerService > - > > Key: FLINK-8322 > URL: https://issues.apache.org/jira/browse/FLINK-8322 > Project: Flink > Issue Type: Improvement > Components: DataStream API >Affects Versions: 1.4.0 >Reporter: Bowen Li >Assignee: Bowen Li > Fix For: 1.5.0 > > > There are pretty common use cases where users want to use timers as scheduled > threads - e.g. add a timer to wake up x hours later and do something (reap > old data usually) only if there's no existing timers, basically we only want > at most 1 timer exists for the key all the time -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5236: [FLINK-8322] support getting number of existing ti...
Github user bowenli86 closed the pull request at: https://github.com/apache/flink/pull/5236 ---
[jira] [Resolved] (FLINK-7049) TestingApplicationMaster keeps running after integration tests finish
[ https://issues.apache.org/jira/browse/FLINK-7049?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Ted Yu resolved FLINK-7049. --- Resolution: Cannot Reproduce > TestingApplicationMaster keeps running after integration tests finish > - > > Key: FLINK-7049 > URL: https://issues.apache.org/jira/browse/FLINK-7049 > Project: Flink > Issue Type: Test > Components: Tests, YARN >Reporter: Ted Yu >Priority: Minor > Attachments: testingApplicationMaster.stack > > > After integration tests finish, TestingApplicationMaster is still running. > Toward the end of > flink-yarn-tests/target/flink-yarn-tests-ha/flink-yarn-tests-ha-logDir-nm-1_0/application_1498768839874_0001/container_1498768839874_0001_03_01/jobmanager.log > : > {code} > 2017-06-29 22:09:49,681 INFO org.apache.zookeeper.ClientCnxn > - Opening socket connection to server 127.0.0.1/127.0.0.1:46165 > 2017-06-29 22:09:49,681 ERROR > org.apache.flink.shaded.org.apache.curator.ConnectionState- > Authentication failed > 2017-06-29 22:09:49,682 WARN org.apache.zookeeper.ClientCnxn > - Session 0x0 for server null, unexpected error, closing socket > connection and attempting reconnect > java.net.ConnectException: Connection refused > at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method) > at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:717) > at > org.apache.zookeeper.ClientCnxnSocketNIO.doTransport(ClientCnxnSocketNIO.java:361) > at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1081) > 2017-06-29 22:09:50,782 WARN org.apache.zookeeper.ClientCnxn > - SASL configuration failed: > javax.security.auth.login.LoginException: No JAAS configuration section named > 'Client' was found in specified JAAS configuration file: > '/tmp/jaas-3597644653611245612.conf'. Will continue connection to Zookeeper > server without SASL authentication, if Zookeeper server allows it. > 2017-06-29 22:09:50,782 INFO org.apache.zookeeper.ClientCnxn > - Opening socket connection to server 127.0.0.1/127.0.0.1:46165 > 2017-06-29 22:09:50,782 ERROR > org.apache.flink.shaded.org.apache.curator.ConnectionState- > Authentication failed > 2017-06-29 22:09:50,783 WARN org.apache.zookeeper.ClientCnxn > - Session 0x0 for server null, unexpected error, closing socket > connection and attempting reconnect > java.net.ConnectException: Connection refused > at sun.nio.ch.SocketChannelImpl.checkConnect(Native Method) > at sun.nio.ch.SocketChannelImpl.finishConnect(SocketChannelImpl.java:717) > at > org.apache.zookeeper.ClientCnxnSocketNIO.doTransport(ClientCnxnSocketNIO.java:361) > at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1081) > {code} -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8430) Implement stream-stream non-window full outer join
Hequn Cheng created FLINK-8430: -- Summary: Implement stream-stream non-window full outer join Key: FLINK-8430 URL: https://issues.apache.org/jira/browse/FLINK-8430 Project: Flink Issue Type: Sub-task Components: Table API & SQL Reporter: Hequn Cheng Assignee: Hequn Cheng -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8429) Implement stream-stream non-window right outer join
Hequn Cheng created FLINK-8429: -- Summary: Implement stream-stream non-window right outer join Key: FLINK-8429 URL: https://issues.apache.org/jira/browse/FLINK-8429 Project: Flink Issue Type: Sub-task Reporter: Hequn Cheng Assignee: Hequn Cheng -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8428) Implement stream-stream non-window left outer join
Hequn Cheng created FLINK-8428: -- Summary: Implement stream-stream non-window left outer join Key: FLINK-8428 URL: https://issues.apache.org/jira/browse/FLINK-8428 Project: Flink Issue Type: Sub-task Reporter: Hequn Cheng Assignee: Hequn Cheng -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-5878) Add stream-stream non-window inner/outer join
[ https://issues.apache.org/jira/browse/FLINK-5878?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-5878: --- Summary: Add stream-stream non-window inner/outer join (was: Add stream-stream inner/outer join) > Add stream-stream non-window inner/outer join > - > > Key: FLINK-5878 > URL: https://issues.apache.org/jira/browse/FLINK-5878 > Project: Flink > Issue Type: New Feature > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > > This task is intended to support stream-stream inner/outer join on > tableAPI/SQL. A brief design doc about inner join is created: > https://docs.google.com/document/d/10oJDw-P9fImD5tc3Stwr7aGIhvem4l8uB2bjLzK72u4/edit > We propose to use the mapState as the backend state interface for this "join" > operator, so this task requires FLINK-4856. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-6094) Implement stream-stream non-window inner join
[ https://issues.apache.org/jira/browse/FLINK-6094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-6094: --- Description: This includes: 1.Implement stream-stream non-window inner join was: This includes: 1.Implement stream-stream proctime non-window inner join 2.Implement the retract process logic for join > Implement stream-stream non-window inner join > -- > > Key: FLINK-6094 > URL: https://issues.apache.org/jira/browse/FLINK-6094 > Project: Flink > Issue Type: Sub-task > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > Fix For: 1.5.0 > > > This includes: > 1.Implement stream-stream non-window inner join -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-5878) Add stream-stream inner/outer join
[ https://issues.apache.org/jira/browse/FLINK-5878?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-5878: --- Description: This task is intended to support stream-stream inner/outer join on tableAPI/SQL. A brief design doc about inner join is created: https://docs.google.com/document/d/10oJDw-P9fImD5tc3Stwr7aGIhvem4l8uB2bjLzK72u4/edit We propose to use the mapState as the backend state interface for this "join" operator, so this task requires FLINK-4856. was: This task is intended to support stream-stream inner join on tableAPI. A brief design doc is created: https://docs.google.com/document/d/10oJDw-P9fImD5tc3Stwr7aGIhvem4l8uB2bjLzK72u4/edit We propose to use the mapState as the backend state interface for this "join" operator, so this task requires FLINK-4856. > Add stream-stream inner/outer join > -- > > Key: FLINK-5878 > URL: https://issues.apache.org/jira/browse/FLINK-5878 > Project: Flink > Issue Type: New Feature > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > > This task is intended to support stream-stream inner/outer join on > tableAPI/SQL. A brief design doc about inner join is created: > https://docs.google.com/document/d/10oJDw-P9fImD5tc3Stwr7aGIhvem4l8uB2bjLzK72u4/edit > We propose to use the mapState as the backend state interface for this "join" > operator, so this task requires FLINK-4856. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-6094) Implement stream-stream non-window inner join
[ https://issues.apache.org/jira/browse/FLINK-6094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-6094: --- Summary: Implement stream-stream non-window inner join (was: Implement stream-stream proctime non-window inner join) > Implement stream-stream non-window inner join > -- > > Key: FLINK-6094 > URL: https://issues.apache.org/jira/browse/FLINK-6094 > Project: Flink > Issue Type: Sub-task > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > Fix For: 1.5.0 > > > This includes: > 1.Implement stream-stream proctime non-window inner join > 2.Implement the retract process logic for join -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-5878) Add stream-stream inner/outer join
[ https://issues.apache.org/jira/browse/FLINK-5878?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-5878: --- Summary: Add stream-stream inner/outer join (was: Add stream-stream inner/left-out join) > Add stream-stream inner/outer join > -- > > Key: FLINK-5878 > URL: https://issues.apache.org/jira/browse/FLINK-5878 > Project: Flink > Issue Type: New Feature > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > > This task is intended to support stream-stream inner join on tableAPI. A > brief design doc is created: > https://docs.google.com/document/d/10oJDw-P9fImD5tc3Stwr7aGIhvem4l8uB2bjLzK72u4/edit > We propose to use the mapState as the backend state interface for this "join" > operator, so this task requires FLINK-4856. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-6094) Implement stream-stream proctime non-window inner join
[ https://issues.apache.org/jira/browse/FLINK-6094?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Hequn Cheng updated FLINK-6094: --- Issue Type: Sub-task (was: New Feature) Parent: FLINK-5878 > Implement stream-stream proctime non-window inner join > --- > > Key: FLINK-6094 > URL: https://issues.apache.org/jira/browse/FLINK-6094 > Project: Flink > Issue Type: Sub-task > Components: Table API & SQL >Reporter: Shaoxuan Wang >Assignee: Hequn Cheng > Fix For: 1.5.0 > > > This includes: > 1.Implement stream-stream proctime non-window inner join > 2.Implement the retract process logic for join -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-8421) HeapInternalTimerService should reconfigure compatible key / namespace serializers on restore
[ https://issues.apache.org/jira/browse/FLINK-8421?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Tzu-Li (Gordon) Tai updated FLINK-8421: --- Description: The {{HeapInternalTimerService}} still uses simple {{equals}} checks on restored / newly provided serializers for compatibility checks. This should be replaced with the {{TypeSerializer::ensureCompatibility}} checks instead, so that new serializers can be reconfigured. This would entail that the {{TypeSerializerConfiguration}} of the key and namespace serializer in the {{HeapInternalTimerService}} also needs to be written to the raw state. For Flink 1.4.0 release and current master, this is a critical bug since the {{KryoSerializer}} has different default base registrations than before due to FLINK-7420. i.e if the key of a window is serialized using the {{KryoSerializer}} in 1.3.x, the restore would never succeed in 1.4.0. For 1.3.x, this fix would be an improvement, such that the {{HeapInternalTimerService}} restore will make use of serializer reconfiguration. Other remarks: * We need to double check all operators that checkpoint / restore from **raw** state. Apparently, the serializer compatibility checks were only implemented for managed state. * Migration ITCases apparently do not have enough coverage. A migration test job that uses a key type which required the {{KryoSerializer}}, and uses windows, would have caught this issue. was: The {{HeapInternalTimerService}} still uses simple {{equals}} checks on restored / newly provided serializers for compatibility checks. This should be replaced with the {{TypeSerializer::ensureCompatibility}} checks instead, so that new serializers can be reconfigured. For Flink 1.4.0 release and current master, this is a critical bug since the {{KryoSerializer}} has different default base registrations than before due to FLINK-7420. i.e if the key of a window is serialized using the {{KryoSerializer}} in 1.3.x, the restore would never succeed in 1.4.0. For 1.3.x, this fix would be an improvement, such that the {{HeapInternalTimerService}} restore will make use of serializer reconfiguration. Other remarks: * We need to double check all operators that checkpoint / restore from **raw** state. Apparently, the serializer compatibility checks were only implemented for managed state. * Migration ITCases apparently do not have enough coverage. A migration test job that uses a key type which required the {{KryoSerializer}}, and uses windows, would have caught this issue. > HeapInternalTimerService should reconfigure compatible key / namespace > serializers on restore > - > > Key: FLINK-8421 > URL: https://issues.apache.org/jira/browse/FLINK-8421 > Project: Flink > Issue Type: Bug >Affects Versions: 1.4.0, 1.5.0 >Reporter: Tzu-Li (Gordon) Tai >Priority: Blocker > Fix For: 1.3.3, 1.5.0, 1.4.1 > > > The {{HeapInternalTimerService}} still uses simple {{equals}} checks on > restored / newly provided serializers for compatibility checks. This should > be replaced with the {{TypeSerializer::ensureCompatibility}} checks instead, > so that new serializers can be reconfigured. > This would entail that the {{TypeSerializerConfiguration}} of the key and > namespace serializer in the {{HeapInternalTimerService}} also needs to be > written to the raw state. > For Flink 1.4.0 release and current master, this is a critical bug since the > {{KryoSerializer}} has different default base registrations than before due > to FLINK-7420. i.e if the key of a window is serialized using the > {{KryoSerializer}} in 1.3.x, the restore would never succeed in 1.4.0. > For 1.3.x, this fix would be an improvement, such that the > {{HeapInternalTimerService}} restore will make use of serializer > reconfiguration. > Other remarks: > * We need to double check all operators that checkpoint / restore from > **raw** state. Apparently, the serializer compatibility checks were only > implemented for managed state. > * Migration ITCases apparently do not have enough coverage. A migration test > job that uses a key type which required the {{KryoSerializer}}, and uses > windows, would have caught this issue. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8340) Do not pass Configuration and configuration directory to CustomCommandLine methods
[ https://issues.apache.org/jira/browse/FLINK-8340?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8340. Resolution: Fixed Fixed via 30011b9b110aad0e1c28e7e0a025b73986781a72 > Do not pass Configuration and configuration directory to CustomCommandLine > methods > -- > > Key: FLINK-8340 > URL: https://issues.apache.org/jira/browse/FLINK-8340 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Currently, all methods in {{CustomCommandLine}} need a {{Configuration}} and > sometimes the configuration directory. Since these values should not change > over the lifetime of the {{CustomCommandLine}} we should pass them as a > constructor argument instead of a method argument. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8341) Remove unneeded CommandLineOptions
[ https://issues.apache.org/jira/browse/FLINK-8341?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8341. Resolution: Fixed Fixed via d7e9dc1931f1f1cedbfee12aebe34dd76e9bac10 > Remove unneeded CommandLineOptions > -- > > Key: FLINK-8341 > URL: https://issues.apache.org/jira/browse/FLINK-8341 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > With the refactorings of the {{CliFrontend}} we no longer have to keep the > JobManager address and commandLine in the {{CommandLineOptions}}. Therefore, > these fields should be removed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8349) Remove Yarn specific commands from YarnClusterClient
[ https://issues.apache.org/jira/browse/FLINK-8349?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8349. Resolution: Fixed Fixed via 402499f06a4b590ac47df64ecc01055c06b0399b > Remove Yarn specific commands from YarnClusterClient > > > Key: FLINK-8349 > URL: https://issues.apache.org/jira/browse/FLINK-8349 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{YarnClusterClient}} should no longer use Yarn specific commands. This > is necessary to make the {{FlinkYarnSessionCli}} work with other > {{ClusterClient}} implementations than the {{YarnClusterClient}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8342) Remove ClusterClient generic type parameter from ClusterDescriptor
[ https://issues.apache.org/jira/browse/FLINK-8342?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8342. Resolution: Fixed Fixed via 10e900b25ac03876d3f9e78f260d48efe6b9d853 > Remove ClusterClient generic type parameter from ClusterDescriptor > -- > > Key: FLINK-8342 > URL: https://issues.apache.org/jira/browse/FLINK-8342 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{ClusterDescriptor}} should not specialize the returned > {{ClusterClient}} type in order to develop code which can work with all > {{ClusterDescriptors}} and {{ClusterClients}}. Therefore, I propose to remove > the generic type parameter from {{ClusterDescriptor}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8347) Make Cluster id typesafe
[ https://issues.apache.org/jira/browse/FLINK-8347?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8347. Resolution: Fixed Fixed via 38d3720863c6187153174d0df57fc414b0cf8e96 > Make Cluster id typesafe > > > Key: FLINK-8347 > URL: https://issues.apache.org/jira/browse/FLINK-8347 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Currently, the cluster id is of type {{String}}. We should make the id > typesafe to avoid mixups between different {{CustomCommandLines}} and > {{ClusterDescriptors}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8420) Timeout exceptions are not properly recognized by RetryingRegistration
[ https://issues.apache.org/jira/browse/FLINK-8420?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324647#comment-16324647 ] ASF GitHub Bot commented on FLINK-8420: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5286 > Timeout exceptions are not properly recognized by RetryingRegistration > -- > > Key: FLINK-8420 > URL: https://issues.apache.org/jira/browse/FLINK-8420 > Project: Flink > Issue Type: Bug > Components: Distributed Coordination >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{RetryingRegistration}} does not correctly respond to > {{TimeoutExceptions}} and instead treats them like errors. This causes that > it waits for the delay on error instead of backing exponentially off. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Resolved] (FLINK-8119) Cannot submit jobs to YARN Session in FLIP-6 mode
[ https://issues.apache.org/jira/browse/FLINK-8119?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann resolved FLINK-8119. -- Resolution: Fixed Fixed via dbe0e8286d76a5facdb49589b638b87dbde80178 > Cannot submit jobs to YARN Session in FLIP-6 mode > - > > Key: FLINK-8119 > URL: https://issues.apache.org/jira/browse/FLINK-8119 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Gary Yao >Assignee: Till Rohrmann >Priority: Blocker > Labels: flip-6 > Fix For: 1.5.0 > > > Cannot submit jobs to YARN Session in FLIP-6 mode because > {{FlinkYarnSessionCli}} becomes the _active_ CLI (should be > {{Flip6DefaultCLI}}). > *Steps to reproduce* > # Build Flink 1.5 {{101fef7397128b0aba23221481ab86f62b18118f}} > # {{bin/yarn-session.sh -flip6 -d -n 1 -s 1 -jm 1024 -tm 1024}} > # {{bin/flink run -flip6 ./examples/streaming/WordCount.jar}} > # Verify that the job will not run. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8347) Make Cluster id typesafe
[ https://issues.apache.org/jira/browse/FLINK-8347?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324650#comment-16324650 ] ASF GitHub Bot commented on FLINK-8347: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5232 > Make Cluster id typesafe > > > Key: FLINK-8347 > URL: https://issues.apache.org/jira/browse/FLINK-8347 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Currently, the cluster id is of type {{String}}. We should make the id > typesafe to avoid mixups between different {{CustomCommandLines}} and > {{ClusterDescriptors}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8119) Cannot submit jobs to YARN Session in FLIP-6 mode
[ https://issues.apache.org/jira/browse/FLINK-8119?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324653#comment-16324653 ] ASF GitHub Bot commented on FLINK-8119: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5234 > Cannot submit jobs to YARN Session in FLIP-6 mode > - > > Key: FLINK-8119 > URL: https://issues.apache.org/jira/browse/FLINK-8119 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Gary Yao >Assignee: Till Rohrmann >Priority: Blocker > Labels: flip-6 > Fix For: 1.5.0 > > > Cannot submit jobs to YARN Session in FLIP-6 mode because > {{FlinkYarnSessionCli}} becomes the _active_ CLI (should be > {{Flip6DefaultCLI}}). > *Steps to reproduce* > # Build Flink 1.5 {{101fef7397128b0aba23221481ab86f62b18118f}} > # {{bin/yarn-session.sh -flip6 -d -n 1 -s 1 -jm 1024 -tm 1024}} > # {{bin/flink run -flip6 ./examples/streaming/WordCount.jar}} > # Verify that the job will not run. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8348) Print help for DefaultCLI
[ https://issues.apache.org/jira/browse/FLINK-8348?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324652#comment-16324652 ] ASF GitHub Bot commented on FLINK-8348: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5233 > Print help for DefaultCLI > - > > Key: FLINK-8348 > URL: https://issues.apache.org/jira/browse/FLINK-8348 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Print help for the {{DefaultCLI}} when calling {{flink -help}}, for example. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8341) Remove unneeded CommandLineOptions
[ https://issues.apache.org/jira/browse/FLINK-8341?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324648#comment-16324648 ] ASF GitHub Bot commented on FLINK-8341: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5227 > Remove unneeded CommandLineOptions > -- > > Key: FLINK-8341 > URL: https://issues.apache.org/jira/browse/FLINK-8341 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > With the refactorings of the {{CliFrontend}} we no longer have to keep the > JobManager address and commandLine in the {{CommandLineOptions}}. Therefore, > these fields should be removed. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8349) Remove Yarn specific commands from YarnClusterClient
[ https://issues.apache.org/jira/browse/FLINK-8349?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324656#comment-16324656 ] ASF GitHub Bot commented on FLINK-8349: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5229 > Remove Yarn specific commands from YarnClusterClient > > > Key: FLINK-8349 > URL: https://issues.apache.org/jira/browse/FLINK-8349 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{YarnClusterClient}} should no longer use Yarn specific commands. This > is necessary to make the {{FlinkYarnSessionCli}} work with other > {{ClusterClient}} implementations than the {{YarnClusterClient}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8342) Remove ClusterClient generic type parameter from ClusterDescriptor
[ https://issues.apache.org/jira/browse/FLINK-8342?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324649#comment-16324649 ] ASF GitHub Bot commented on FLINK-8342: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5228 > Remove ClusterClient generic type parameter from ClusterDescriptor > -- > > Key: FLINK-8342 > URL: https://issues.apache.org/jira/browse/FLINK-8342 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{ClusterDescriptor}} should not specialize the returned > {{ClusterClient}} type in order to develop code which can work with all > {{ClusterDescriptors}} and {{ClusterClients}}. Therefore, I propose to remove > the generic type parameter from {{ClusterDescriptor}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8201) YarnResourceManagerTest causes license checking failure
[ https://issues.apache.org/jira/browse/FLINK-8201?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324654#comment-16324654 ] ASF GitHub Bot commented on FLINK-8201: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5123 > YarnResourceManagerTest causes license checking failure > --- > > Key: FLINK-8201 > URL: https://issues.apache.org/jira/browse/FLINK-8201 > Project: Flink > Issue Type: Bug >Reporter: Wenlong Lyu >Assignee: Wenlong Lyu > Fix For: 1.5.0 > > > YarnResourceManagerTest generates a temporary taskmanager config file in > flink-yarn module > root folder and never clear it, which makes license checking fail when we > run {{mvn clean verify}} multiple times in the same source folder. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Closed] (FLINK-8348) Print help for DefaultCLI
[ https://issues.apache.org/jira/browse/FLINK-8348?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8348. Resolution: Feedback Received Fixed via 7d986ce0482562efd90fec416c4b2f4638a4058a > Print help for DefaultCLI > - > > Key: FLINK-8348 > URL: https://issues.apache.org/jira/browse/FLINK-8348 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Print help for the {{DefaultCLI}} when calling {{flink -help}}, for example. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8340) Do not pass Configuration and configuration directory to CustomCommandLine methods
[ https://issues.apache.org/jira/browse/FLINK-8340?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324655#comment-16324655 ] ASF GitHub Bot commented on FLINK-8340: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5226 > Do not pass Configuration and configuration directory to CustomCommandLine > methods > -- > > Key: FLINK-8340 > URL: https://issues.apache.org/jira/browse/FLINK-8340 > Project: Flink > Issue Type: Sub-task > Components: Client >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > Currently, all methods in {{CustomCommandLine}} need a {{Configuration}} and > sometimes the configuration directory. Since these values should not change > over the lifetime of the {{CustomCommandLine}} we should pass them as a > constructor argument instead of a method argument. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8288) Register the web interface url to yarn for yarn job mode
[ https://issues.apache.org/jira/browse/FLINK-8288?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324651#comment-16324651 ] ASF GitHub Bot commented on FLINK-8288: --- Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5186 > Register the web interface url to yarn for yarn job mode > > > Key: FLINK-8288 > URL: https://issues.apache.org/jira/browse/FLINK-8288 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > Fix For: 1.5.0 > > > For flip-6 job mode, the resource manager is created before the web monitor, > so the web interface url is not set to resource manager, and the resource > manager can not register the url to yarn. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5227: [FLINK-8341] [flip6] Remove not needed options fro...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5227 ---
[GitHub] flink pull request #5229: [FLINK-8349] [flip6] Remove Yarn specific commands...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5229 ---
[GitHub] flink pull request #5233: [FLINK-8348] [flip6] Print help for DefaultCLI
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5233 ---
[GitHub] flink pull request #5228: [FLINK-8342] [flip6] Remove generic type parameter...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5228 ---
[GitHub] flink pull request #5226: [FLINK-8340] [flip6] Remove passing of Configurati...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5226 ---
[jira] [Closed] (FLINK-8420) Timeout exceptions are not properly recognized by RetryingRegistration
[ https://issues.apache.org/jira/browse/FLINK-8420?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann closed FLINK-8420. Resolution: Fixed Fixed via 3c99ae8959f69325bb9b7d810b41c60e42e602c5 > Timeout exceptions are not properly recognized by RetryingRegistration > -- > > Key: FLINK-8420 > URL: https://issues.apache.org/jira/browse/FLINK-8420 > Project: Flink > Issue Type: Bug > Components: Distributed Coordination >Affects Versions: 1.5.0 >Reporter: Till Rohrmann >Assignee: Till Rohrmann > Labels: flip-6 > Fix For: 1.5.0 > > > The {{RetryingRegistration}} does not correctly respond to > {{TimeoutExceptions}} and instead treats them like errors. This causes that > it waits for the delay on error instead of backing exponentially off. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Resolved] (FLINK-8288) Register the web interface url to yarn for yarn job mode
[ https://issues.apache.org/jira/browse/FLINK-8288?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann resolved FLINK-8288. -- Resolution: Fixed Fix Version/s: 1.5.0 Fixed via 3b01686851e8281642924c4a620fb43b008de174 > Register the web interface url to yarn for yarn job mode > > > Key: FLINK-8288 > URL: https://issues.apache.org/jira/browse/FLINK-8288 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > Fix For: 1.5.0 > > > For flip-6 job mode, the resource manager is created before the web monitor, > so the web interface url is not set to resource manager, and the resource > manager can not register the url to yarn. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5186: [FLINK-8288] [runtime] register job master rest en...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5186 ---
[GitHub] flink pull request #5234: [FLINK-8119] [flip6] Wire correct Flip6 components...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5234 ---
[GitHub] flink pull request #5286: [FLINK-8420] [flip6] Recognize TimeoutException in...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5286 ---
[jira] [Resolved] (FLINK-8201) YarnResourceManagerTest causes license checking failure
[ https://issues.apache.org/jira/browse/FLINK-8201?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Till Rohrmann resolved FLINK-8201. -- Resolution: Fixed Fix Version/s: 1.5.0 Fixed via d0bc300087b037b156425ccd509147177dd9529e > YarnResourceManagerTest causes license checking failure > --- > > Key: FLINK-8201 > URL: https://issues.apache.org/jira/browse/FLINK-8201 > Project: Flink > Issue Type: Bug >Reporter: Wenlong Lyu >Assignee: Wenlong Lyu > Fix For: 1.5.0 > > > YarnResourceManagerTest generates a temporary taskmanager config file in > flink-yarn module > root folder and never clear it, which makes license checking fail when we > run {{mvn clean verify}} multiple times in the same source folder. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5232: [FLINK-8347] [flip6] Make cluster id used by Clust...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5232 ---
[GitHub] flink pull request #5123: [FLINK-8201] YarnResourceManagerTest causes licens...
Github user asfgit closed the pull request at: https://github.com/apache/flink/pull/5123 ---
[jira] [Created] (FLINK-8427) Checkstyle for org.apache.flink.optimizer.costs
Greg Hogan created FLINK-8427: - Summary: Checkstyle for org.apache.flink.optimizer.costs Key: FLINK-8427 URL: https://issues.apache.org/jira/browse/FLINK-8427 Project: Flink Issue Type: Improvement Components: Optimizer Affects Versions: 1.5.0 Reporter: Greg Hogan Assignee: Greg Hogan Priority: Trivial -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8414) Gelly performance seriously decreases when using the suggested parallelism configuration
[ https://issues.apache.org/jira/browse/FLINK-8414?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324542#comment-16324542 ] Greg Hogan commented on FLINK-8414: --- It is incumbent on the user to configure an appropriate parallelism for the quantity of data. Those graphs contain only a few tens of megabytes of data so it is not surprising that the optimal parallelism is around (or even lower than) 16. You can use `VertexMetrics` to pre-compute the size of the graph and adjust the parallelism at runtime (`ExecutionConfig#setParallelism`). Flink and Gelly are designed to scale to 100s to 1000s of parallel tasks and GBs to TBs of data. > Gelly performance seriously decreases when using the suggested parallelism > configuration > > > Key: FLINK-8414 > URL: https://issues.apache.org/jira/browse/FLINK-8414 > Project: Flink > Issue Type: Bug > Components: Configuration, Documentation, Gelly >Reporter: flora karniav >Priority: Minor > > I am running Gelly examples with different datasets in a cluster of 5 > machines (1 Jobmanager and 4 Taskmanagers) of 32 cores each. > The number of Slots parameter is set to 32 (as suggested) and the parallelism > to 128 (32 cores*4 taskmanagers). > I observe a vast performance degradation using these suggested settings than > setting parallelism.default to 16 for example were the same job completes at > ~60 seconds vs ~140 in the 128 parallelism case. > Is there something wrong in my configuration? Should I decrease parallelism > and -if so- will this inevitably decrease CPU utilization? > Another matter that may be related to this is the number of partitions of the > data. Is this somehow related to parallelism? How many partitions are created > in the case of parallelism.default=128? -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-8426) Error in Generating Timestamp/Watermakr doc
[ https://issues.apache.org/jira/browse/FLINK-8426?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Christophe Jolif updated FLINK-8426: Description: In https://ci.apache.org/projects/flink/flink-docs-release-1.4/dev/event_timestamps_watermarks.html {{public class BoundedOutOfOrdernessGenerator extends AssignerWithPeriodicWatermarks}} should be {{public class BoundedOutOfOrdernessGenerator implements AssignerWithPeriodicWatermarks}} was: In https://ci.apache.org/projects/flink/flink-docs-release-1.4/dev/event_timestamps_watermarks.html {{ public class BoundedOutOfOrdernessGenerator extends AssignerWithPeriodicWatermarks}} should be {{ public class BoundedOutOfOrdernessGenerator implements AssignerWithPeriodicWatermarks}} > Error in Generating Timestamp/Watermakr doc > --- > > Key: FLINK-8426 > URL: https://issues.apache.org/jira/browse/FLINK-8426 > Project: Flink > Issue Type: Bug > Components: Documentation >Affects Versions: 1.4.0 >Reporter: Christophe Jolif >Priority: Trivial > > In > https://ci.apache.org/projects/flink/flink-docs-release-1.4/dev/event_timestamps_watermarks.html > {{public class BoundedOutOfOrdernessGenerator extends > AssignerWithPeriodicWatermarks}} > should be > {{public class BoundedOutOfOrdernessGenerator implements > AssignerWithPeriodicWatermarks}} -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8426) Error in Generating Timestamp/Watermakr doc
Christophe Jolif created FLINK-8426: --- Summary: Error in Generating Timestamp/Watermakr doc Key: FLINK-8426 URL: https://issues.apache.org/jira/browse/FLINK-8426 Project: Flink Issue Type: Bug Components: Documentation Affects Versions: 1.4.0 Reporter: Christophe Jolif Priority: Trivial In https://ci.apache.org/projects/flink/flink-docs-release-1.4/dev/event_timestamps_watermarks.html {{ public class BoundedOutOfOrdernessGenerator extends AssignerWithPeriodicWatermarks}} should be {{ public class BoundedOutOfOrdernessGenerator implements AssignerWithPeriodicWatermarks}} -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8422) Checkstyle for org.apache.flink.api.java.tuple
[ https://issues.apache.org/jira/browse/FLINK-8422?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324528#comment-16324528 ] ASF GitHub Bot commented on FLINK-8422: --- GitHub user greghogan opened a pull request: https://github.com/apache/flink/pull/5292 [FLINK-8422] [core] Checkstyle for org.apache.flink.api.java.tuple ## What is the purpose of the change Update TupleGenerator for Flink's checkstyle and rebuild Tuple and TupleBuilder classes. ## Brief change log `TupleGenerator` has been updated to write Flink-checkstyle compliant code. The following non-generated files were manually updated: `Tuple`, `Tuple0`, `Tuple0Builder`, `Tuple2Test` `org.apache.flink.api.java.tuple` was removed from the checkstyle suppressions for `flink-core`. ## Verifying this change This change is a trivial rework / code cleanup without any test coverage. Re-running `TupleGenerator` should replicate the newly updated files. You can merge this pull request into a Git repository by running: $ git pull https://github.com/greghogan/flink 8422_checkstyle_for_org_apache_flink_api_java_tuple Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5292.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5292 commit b2fa893e20ed09b3898c64f6821f2d177369f0a7 Author: Greg HoganDate: 2018-01-12T16:35:15Z [FLINK-8422] [core] Checkstyle for org.apache.flink.api.java.tuple Update TupleGenerator for Flink's checkstyle and rebuild Tuple and TupleBuilder classes. > Checkstyle for org.apache.flink.api.java.tuple > -- > > Key: FLINK-8422 > URL: https://issues.apache.org/jira/browse/FLINK-8422 > Project: Flink > Issue Type: Improvement > Components: Core >Affects Versions: 1.5.0 >Reporter: Greg Hogan >Assignee: Greg Hogan >Priority: Trivial > > Update {{TupleGenerator}} for Flink's checkstyle and rebuild {{Tuple}} and > {{TupleBuilder}} classes. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5292: [FLINK-8422] [core] Checkstyle for org.apache.flin...
GitHub user greghogan opened a pull request: https://github.com/apache/flink/pull/5292 [FLINK-8422] [core] Checkstyle for org.apache.flink.api.java.tuple ## What is the purpose of the change Update TupleGenerator for Flink's checkstyle and rebuild Tuple and TupleBuilder classes. ## Brief change log `TupleGenerator` has been updated to write Flink-checkstyle compliant code. The following non-generated files were manually updated: `Tuple`, `Tuple0`, `Tuple0Builder`, `Tuple2Test` `org.apache.flink.api.java.tuple` was removed from the checkstyle suppressions for `flink-core`. ## Verifying this change This change is a trivial rework / code cleanup without any test coverage. Re-running `TupleGenerator` should replicate the newly updated files. You can merge this pull request into a Git repository by running: $ git pull https://github.com/greghogan/flink 8422_checkstyle_for_org_apache_flink_api_java_tuple Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5292.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5292 commit b2fa893e20ed09b3898c64f6821f2d177369f0a7 Author: Greg HoganDate: 2018-01-12T16:35:15Z [FLINK-8422] [core] Checkstyle for org.apache.flink.api.java.tuple Update TupleGenerator for Flink's checkstyle and rebuild Tuple and TupleBuilder classes. ---
[jira] [Commented] (FLINK-8361) Remove create_release_files.sh
[ https://issues.apache.org/jira/browse/FLINK-8361?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324525#comment-16324525 ] ASF GitHub Bot commented on FLINK-8361: --- GitHub user greghogan opened a pull request: https://github.com/apache/flink/pull/5291 [FLINK-8361] [build] Remove create_release_files.sh ## What is the purpose of the change The monolithic create_release_files.sh does not support building without Hadoop and has been superseded by the scripts in tools/releasing. You can merge this pull request into a Git repository by running: $ git pull https://github.com/greghogan/flink 8361_remove_create_release_files_sh Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5291.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5291 commit d9af86339ba592161419be25b6046ccfa4b47b2f Author: Greg HoganDate: 2018-01-12T15:44:13Z [FLINK-8361] [build] Remove create_release_files.sh The monolithic create_release_files.sh does not support building without Hadoop and has been superseded by the scripts in tools/releasing. > Remove create_release_files.sh > -- > > Key: FLINK-8361 > URL: https://issues.apache.org/jira/browse/FLINK-8361 > Project: Flink > Issue Type: Improvement > Components: Build System >Affects Versions: 1.5.0 >Reporter: Greg Hogan >Priority: Trivial > > The monolithic {{create_release_files.sh}} does not support building Flink > without Hadoop and looks to have been superseded by the scripts in > {{tools/releasing}}. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5291: [FLINK-8361] [build] Remove create_release_files.s...
GitHub user greghogan opened a pull request: https://github.com/apache/flink/pull/5291 [FLINK-8361] [build] Remove create_release_files.sh ## What is the purpose of the change The monolithic create_release_files.sh does not support building without Hadoop and has been superseded by the scripts in tools/releasing. You can merge this pull request into a Git repository by running: $ git pull https://github.com/greghogan/flink 8361_remove_create_release_files_sh Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5291.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5291 commit d9af86339ba592161419be25b6046ccfa4b47b2f Author: Greg HoganDate: 2018-01-12T15:44:13Z [FLINK-8361] [build] Remove create_release_files.sh The monolithic create_release_files.sh does not support building without Hadoop and has been superseded by the scripts in tools/releasing. ---
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324469#comment-16324469 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161248402 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/ResultSubpartitionView.java --- @@ -52,4 +52,9 @@ boolean isReleased(); Throwable getFailureCause(); + + /** +* Returns whether the next buffer is event or not. --- End diff -- `is an event` > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324471#comment-16324471 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161296711 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpilledSubpartitionView.java --- @@ -73,6 +73,9 @@ /** Flag indicating whether all resources have been released. */ private AtomicBoolean isReleased = new AtomicBoolean(); + /** The next buffer to hand out. */ + private Buffer nextBuffer; --- End diff -- We need to protect this against race conditions with respect to `releaseAllResources()` as well. Actually, I'm surprised that nothing in this class is protected against concurrently releasing it. Although I have created a separate issue for this ([FLINK-8425](https://issues.apache.org/jira/browse/FLINK-8425)), I think we may need to solve this here for the new `nextBuffer` anyway. > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324467#comment-16324467 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161295951 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpillableSubpartitionView.java --- @@ -199,6 +199,19 @@ public boolean isReleased() { } } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } + + if (spilledView != null) { --- End diff -- `checkState(spilledView != null, "No in-memory buffers available, but also nothing spilled.");` just like in `getNextBuffer()`? > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324472#comment-16324472 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161249082 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpillableSubpartitionView.java --- @@ -199,6 +199,19 @@ public boolean isReleased() { } } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } --- End diff -- We probably need synchronization here to access `nextBuffer` and also check for `isReleased()` similar to `getNextBuffer`, since we are basically doing the same but without taking the buffer. Why not integrate this into `getNextBuffer` and the `BufferAndBacklog` returned there? Inside that method, gathering this additional info is basically for free and we may thus speed up some code paths. > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324466#comment-16324466 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161295428 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpilledSubpartitionView.java --- @@ -163,6 +190,15 @@ public boolean isReleased() { return parent.isReleased() || isReleased.get(); } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } + + return false; --- End diff -- I'm afraid relying on `nextBuffer` won't be enough because it might not have been there during `getNextBuffer()` but may be there now. Please add a test for this case. > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324465#comment-16324465 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161247589 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- how about checking `numCreditsAvailable` first since that's the cheaper check? > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324473#comment-16324473 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161287607 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- one more thing: in the special case of using it inside `getNextBuffer()`, we already retrieved the `remaining` number of buffers so we can spare one lookup into an atomic integer here by passing this one in > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324470#comment-16324470 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161282684 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- also, why not simply do the following for this whole method? ``` return numBuffersAvailable.get() > 0 && (numCreditsAvailable > 0 || subpartitionView.nextBufferIsEvent()); ``` > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324464#comment-16324464 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161241148 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -49,10 +50,24 @@ private volatile ResultSubpartitionView subpartitionView; + /** +* The status indicating whether this reader is already enqueued in the pipeline for transferring +* data or not. It is mainly used for avoid registering this reader to the pipeline repeatedly. +*/ + private boolean isRegisteredAvailable; + + /** The number of available buffers for holding data on the consumer side. */ + private int numCreditsAvailable; --- End diff -- Just a note since I was wondering whether we need synchronization here (not needed after verifying the things below): 1) `numCreditsAvailable` is increased via `PartitionRequestServerHandler#channelRead0` which is a separate channel handler than `PartitionRequestQueue` (see `NettyProtocol#getServerChannelHandlers`). According to [Netty's thread model](https://netty.io/wiki/new-and-noteworthy-in-4.0.html#wiki-h2-34), we should be safe though: > A user can specify an EventExecutor when he or she adds a handler to a ChannelPipeline. > - If specified, the handler methods of the ChannelHandler are always invoked by the specified EventExecutor. > - If unspecified, the handler methods are always invoked by the EventLoop that its associated Channel is registered to. 2) `numCreditsAvailable` is read from `PartitionRequestQueue#enqueueAvailableReader()` and `SequenceNumberingViewReader#getNextBuffer()` which are both accessed by the channel's IO thread only. > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-7456) Implement Netty sender incoming pipeline for credit-based
[ https://issues.apache.org/jira/browse/FLINK-7456?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324468#comment-16324468 ] ASF GitHub Bot commented on FLINK-7456: --- Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161242036 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -49,10 +50,24 @@ private volatile ResultSubpartitionView subpartitionView; + /** +* The status indicating whether this reader is already enqueued in the pipeline for transferring +* data or not. It is mainly used for avoid registering this reader to the pipeline repeatedly. +*/ + private boolean isRegisteredAvailable; --- End diff -- I know it's default-initialized to false but let's make this explicit > Implement Netty sender incoming pipeline for credit-based > - > > Key: FLINK-7456 > URL: https://issues.apache.org/jira/browse/FLINK-7456 > Project: Flink > Issue Type: Sub-task > Components: Network >Reporter: zhijiang >Assignee: zhijiang > Fix For: 1.5.0 > > > This is a part of work for credit-based network flow control. > On sender side, each subpartition view maintains an atomic integer > {{currentCredit}} from receiver. Once receiving the messages of > {{PartitionRequest}} and {{AddCredit}}, the {{currentCredit}} is added by > deltas. > Each view also maintains an atomic boolean field to mark it as registered > available for transfer to make sure it is enqueued in handler only once. If > the {{currentCredit}} increases from zero and there are available buffers in > the subpartition, the corresponding view will be enqueued for transferring > data. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161248402 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/ResultSubpartitionView.java --- @@ -52,4 +52,9 @@ boolean isReleased(); Throwable getFailureCause(); + + /** +* Returns whether the next buffer is event or not. --- End diff -- `is an event` ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161247589 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- how about checking `numCreditsAvailable` first since that's the cheaper check? ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161241148 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -49,10 +50,24 @@ private volatile ResultSubpartitionView subpartitionView; + /** +* The status indicating whether this reader is already enqueued in the pipeline for transferring +* data or not. It is mainly used for avoid registering this reader to the pipeline repeatedly. +*/ + private boolean isRegisteredAvailable; + + /** The number of available buffers for holding data on the consumer side. */ + private int numCreditsAvailable; --- End diff -- Just a note since I was wondering whether we need synchronization here (not needed after verifying the things below): 1) `numCreditsAvailable` is increased via `PartitionRequestServerHandler#channelRead0` which is a separate channel handler than `PartitionRequestQueue` (see `NettyProtocol#getServerChannelHandlers`). According to [Netty's thread model](https://netty.io/wiki/new-and-noteworthy-in-4.0.html#wiki-h2-34), we should be safe though: > A user can specify an EventExecutor when he or she adds a handler to a ChannelPipeline. > - If specified, the handler methods of the ChannelHandler are always invoked by the specified EventExecutor. > - If unspecified, the handler methods are always invoked by the EventLoop that its associated Channel is registered to. 2) `numCreditsAvailable` is read from `PartitionRequestQueue#enqueueAvailableReader()` and `SequenceNumberingViewReader#getNextBuffer()` which are both accessed by the channel's IO thread only. ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161296711 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpilledSubpartitionView.java --- @@ -73,6 +73,9 @@ /** Flag indicating whether all resources have been released. */ private AtomicBoolean isReleased = new AtomicBoolean(); + /** The next buffer to hand out. */ + private Buffer nextBuffer; --- End diff -- We need to protect this against race conditions with respect to `releaseAllResources()` as well. Actually, I'm surprised that nothing in this class is protected against concurrently releasing it. Although I have created a separate issue for this ([FLINK-8425](https://issues.apache.org/jira/browse/FLINK-8425)), I think we may need to solve this here for the new `nextBuffer` anyway. ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161282684 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- also, why not simply do the following for this whole method? ``` return numBuffersAvailable.get() > 0 && (numCreditsAvailable > 0 || subpartitionView.nextBufferIsEvent()); ``` ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161249082 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpillableSubpartitionView.java --- @@ -199,6 +199,19 @@ public boolean isReleased() { } } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } --- End diff -- We probably need synchronization here to access `nextBuffer` and also check for `isReleased()` similar to `getNextBuffer`, since we are basically doing the same but without taking the buffer. Why not integrate this into `getNextBuffer` and the `BufferAndBacklog` returned there? Inside that method, gathering this additional info is basically for free and we may thus speed up some code paths. ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161295428 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpilledSubpartitionView.java --- @@ -163,6 +190,15 @@ public boolean isReleased() { return parent.isReleased() || isReleased.get(); } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } + + return false; --- End diff -- I'm afraid relying on `nextBuffer` won't be enough because it might not have been there during `getNextBuffer()` but may be there now. Please add a test for this case. ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161242036 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -49,10 +50,24 @@ private volatile ResultSubpartitionView subpartitionView; + /** +* The status indicating whether this reader is already enqueued in the pipeline for transferring +* data or not. It is mainly used for avoid registering this reader to the pipeline repeatedly. +*/ + private boolean isRegisteredAvailable; --- End diff -- I know it's default-initialized to false but let's make this explicit ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161295951 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/partition/SpillableSubpartitionView.java --- @@ -199,6 +199,19 @@ public boolean isReleased() { } } + @Override + public boolean nextBufferIsEvent() { + if (nextBuffer != null) { + return !nextBuffer.isBuffer(); + } + + if (spilledView != null) { --- End diff -- `checkState(spilledView != null, "No in-memory buffers available, but also nothing spilled.");` just like in `getNextBuffer()`? ---
[GitHub] flink pull request #4552: [FLINK-7456][network] Implement Netty sender incom...
Github user NicoK commented on a diff in the pull request: https://github.com/apache/flink/pull/4552#discussion_r161287607 --- Diff: flink-runtime/src/main/java/org/apache/flink/runtime/io/network/netty/SequenceNumberingViewReader.java --- @@ -77,6 +92,46 @@ void requestSubpartitionView( } } + /** +* The credits from consumer are added in incremental way. +* +* @param creditDeltas The credit deltas +*/ + public void addCredit(int creditDeltas) { + numCreditsAvailable += creditDeltas; + } + + /** +* Updates the value to indicate whether the reader is enqueued in the pipeline or not. +* +* @param isRegisteredAvailable True if this reader is already enqueued in the pipeline. +*/ + public void notifyAvailabilityChanged(boolean isRegisteredAvailable) { + this.isRegisteredAvailable = isRegisteredAvailable; + } + + public boolean isRegisteredAvailable() { + return isRegisteredAvailable; + } + + /** +* Check whether this reader is available or not. +* +* Return true only if the next buffer is event or the reader has both available +* credits and buffers. +*/ + public boolean isAvailable() { + if (numBuffersAvailable.get() <= 0) { + return false; + } + + if (subpartitionView.nextBufferIsEvent() || numCreditsAvailable > 0) { --- End diff -- one more thing: in the special case of using it inside `getNextBuffer()`, we already retrieved the `remaining` number of buffers so we can spare one lookup into an atomic integer here by passing this one in ---
[jira] [Commented] (FLINK-7738) Create WebSocket handler (server)
[ https://issues.apache.org/jira/browse/FLINK-7738?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324419#comment-16324419 ] ASF GitHub Bot commented on FLINK-7738: --- Github user EronWright commented on the issue: https://github.com/apache/flink/pull/4767 @tillrohrmann are you still interested in this websocket code for the REST server? Aside from rebasing, any 'must fix' issues here? > Create WebSocket handler (server) > - > > Key: FLINK-7738 > URL: https://issues.apache.org/jira/browse/FLINK-7738 > Project: Flink > Issue Type: Sub-task > Components: Cluster Management, Mesos >Reporter: Eron Wright >Assignee: Eron Wright > > An abstract handler is needed to support websocket communication. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink issue #4767: [FLINK-7738] [flip-6] Create WebSocket handler (server, c...
Github user EronWright commented on the issue: https://github.com/apache/flink/pull/4767 @tillrohrmann are you still interested in this websocket code for the REST server? Aside from rebasing, any 'must fix' issues here? ---
[jira] [Created] (FLINK-8425) SpilledSubpartitionView not protected against concurrent release calls
Nico Kruber created FLINK-8425: -- Summary: SpilledSubpartitionView not protected against concurrent release calls Key: FLINK-8425 URL: https://issues.apache.org/jira/browse/FLINK-8425 Project: Flink Issue Type: Bug Components: Network Affects Versions: 1.3.2, 1.4.0, 1.3.1, 1.2.1, 1.3.0, 1.1.4, 1.1.3, 1.1.2, 1.1.1, 1.2.0, 1.0.3, 1.0.2, 1.0.1, 1.1.0 Reporter: Nico Kruber Priority: Minor It seems like {{SpilledSubpartitionView}} is not protected against concurrently calling {{releaseAllResources}} as the other {{ResultSubpartitionView}} implementations. These may happen due to failures, e.g. network channels breaking, and will probably only result in some unexpected exceptions being thrown, e.g. from reading from a closed file reader. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink issue #5112: [FLINK-8175] [DataStream API java/scala] remove flink-str...
Github user bowenli86 commented on the issue: https://github.com/apache/flink/pull/5112 @StephanEwen let me know if this is good to go ---
[jira] [Commented] (FLINK-8175) remove flink-streaming-contrib and migrate its classes to flink-streaming-java/scala
[ https://issues.apache.org/jira/browse/FLINK-8175?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324323#comment-16324323 ] ASF GitHub Bot commented on FLINK-8175: --- Github user bowenli86 commented on the issue: https://github.com/apache/flink/pull/5112 @StephanEwen let me know if this is good to go > remove flink-streaming-contrib and migrate its classes to > flink-streaming-java/scala > > > Key: FLINK-8175 > URL: https://issues.apache.org/jira/browse/FLINK-8175 > Project: Flink > Issue Type: Sub-task >Affects Versions: 1.5.0 >Reporter: Bowen Li >Assignee: Bowen Li > Fix For: 1.5.0 > > > I propose removing flink-streaming-contrib from flink-contrib, and migrating > its classes to flink-streaming-java/scala for the following reasons: > - flink-streaming-contrib is so small that it only has 4 classes (3 java and > 1 scala), and they don't need a dedicated jar for Flink to distribute and > maintain it and for users to deal with the overhead of dependency management > - the 4 classes in flink-streaming-contrib are logically more tied to > flink-streaming-java/scala, and thus can be easily migrated > - flink-contrib is already too crowded and noisy. It contains lots of sub > modules with different purposes which confuse developers and users, and they > lack a proper project hierarchy > To take a step even forward, I would argue that even flink-contrib should be > removed and all its submodules should be migrated to other top-level modules > for the following reasons: 1) Apache Flink the whole project itself is a > result of contributions from many developers, there's no reason to highlight > some contributions in a dedicated module named 'contrib' 2) flink-contrib > inherently doesn't have a good hierarchy to hold submodules -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-8424) [Cassandra Connector] Update Cassandra version to last one
[ https://issues.apache.org/jira/browse/FLINK-8424?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joao Boto updated FLINK-8424: - Component/s: Cassandra Connector > [Cassandra Connector] Update Cassandra version to last one > -- > > Key: FLINK-8424 > URL: https://issues.apache.org/jira/browse/FLINK-8424 > Project: Flink > Issue Type: Improvement > Components: Cassandra Connector >Reporter: Joao Boto > > Cassandra connector are using a version outdated > This is to upgrade the cassandra version to something new > https://git1-us-west.apache.org/repos/asf?p=cassandra.git;a=blob_plain;f=CHANGES.txt;hb=refs/tags/cassandra-3.11.1 -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Updated] (FLINK-8424) [Cassandra Connector] Update Cassandra version to last one
[ https://issues.apache.org/jira/browse/FLINK-8424?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joao Boto updated FLINK-8424: - Description: Cassandra connector are using a version outdated This is to upgrade the cassandra version to something new https://git1-us-west.apache.org/repos/asf?p=cassandra.git;a=blob_plain;f=CHANGES.txt;hb=refs/tags/cassandra-3.11.1 was: Cassandra connector are using a version release in the beginning of 2016 This is to upgrade the cassandra version to something new > [Cassandra Connector] Update Cassandra version to last one > -- > > Key: FLINK-8424 > URL: https://issues.apache.org/jira/browse/FLINK-8424 > Project: Flink > Issue Type: Improvement >Reporter: Joao Boto > > Cassandra connector are using a version outdated > This is to upgrade the cassandra version to something new > https://git1-us-west.apache.org/repos/asf?p=cassandra.git;a=blob_plain;f=CHANGES.txt;hb=refs/tags/cassandra-3.11.1 -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink issue #5290: [Flink-8424][Cassandra Connector] upgrade of cassandra ve...
Github user eskabetxe commented on the issue: https://github.com/apache/flink/pull/5290 @twalthr @fhueske ---
[GitHub] flink pull request #5290: [Flink-8424][Cassandra Connector] upgrade of cassa...
GitHub user eskabetxe opened a pull request: https://github.com/apache/flink/pull/5290 [Flink-8424][Cassandra Connector] upgrade of cassandra version and driver to latest ## What is the purpose of the change Update Cassandra version and driver to something new, the driver is current in 3 minor versions behind the version is 1 major version behind ## Verifying this change This change is already covered by existing tests, such as *(please describe tests)*. ## Does this pull request potentially affect one of the following parts: - Dependencies (does it add or upgrade a dependency): yes - The public API, i.e., is any changed class annotated with `@Public(Evolving)`: no - The serializers: no - The runtime per-record code paths (performance sensitive): no - Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: no - The S3 file system connector: no ## Documentation - Does this pull request introduce a new feature? no You can merge this pull request into a Git repository by running: $ git pull https://github.com/eskabetxe/flink FLINK-8424 Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5290.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5290 commit 7e7dd921cf9112c1c2174890f460f47529c8dcb4 Author: eskabetxeDate: 2018-01-12T17:37:06Z FLINK-8424: upgrade cassandra driver and version to latest ---
[jira] [Updated] (FLINK-8424) [Cassandra Connector] Update Cassandra version to last one
[ https://issues.apache.org/jira/browse/FLINK-8424?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ] Joao Boto updated FLINK-8424: - Priority: Major (was: Critical) > [Cassandra Connector] Update Cassandra version to last one > -- > > Key: FLINK-8424 > URL: https://issues.apache.org/jira/browse/FLINK-8424 > Project: Flink > Issue Type: Improvement >Reporter: Joao Boto > > Cassandra connector are using a version release in the beginning of 2016 > This is to upgrade the cassandra version to something new -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8424) [Cassandra Connector] Update Cassandra version to last one
Joao Boto created FLINK-8424: Summary: [Cassandra Connector] Update Cassandra version to last one Key: FLINK-8424 URL: https://issues.apache.org/jira/browse/FLINK-8424 Project: Flink Issue Type: Improvement Reporter: Joao Boto Priority: Critical Cassandra connector are using a version release in the beginning of 2016 This is to upgrade the cassandra version to something new -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8414) Gelly performance seriously decreases when using the suggested parallelism configuration
[ https://issues.apache.org/jira/browse/FLINK-8414?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324228#comment-16324228 ] flora karniav commented on FLINK-8414: -- Thank you for your reply, I am running the ConnectedComponents and PageRank algorithms from Gelly examples on two SNAP datasets: 1) https://snap.stanford.edu/data/egonets-Twitter.html - 81,306 vertices and 2,420,766 edges. 2) https://snap.stanford.edu/data/com-Youtube.html - 1,134,890 vertices and 2,987,624 edges. I also want to point out that I looked into CPU utilization when changing the parallelism level and it seems to grow as expected, however performance is still reduced. (I am sorry if I posted in an inappropriate section but thought of the issue bizarre enough to be configuration or bug-related.) > Gelly performance seriously decreases when using the suggested parallelism > configuration > > > Key: FLINK-8414 > URL: https://issues.apache.org/jira/browse/FLINK-8414 > Project: Flink > Issue Type: Bug > Components: Configuration, Documentation, Gelly >Reporter: flora karniav >Priority: Minor > > I am running Gelly examples with different datasets in a cluster of 5 > machines (1 Jobmanager and 4 Taskmanagers) of 32 cores each. > The number of Slots parameter is set to 32 (as suggested) and the parallelism > to 128 (32 cores*4 taskmanagers). > I observe a vast performance degradation using these suggested settings than > setting parallelism.default to 16 for example were the same job completes at > ~60 seconds vs ~140 in the 128 parallelism case. > Is there something wrong in my configuration? Should I decrease parallelism > and -if so- will this inevitably decrease CPU utilization? > Another matter that may be related to this is the number of partitions of the > data. Is this somehow related to parallelism? How many partitions are created > in the case of parallelism.default=128? -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8423) OperatorChain#pushToOperator catch block may fail with NPE
Chesnay Schepler created FLINK-8423: --- Summary: OperatorChain#pushToOperator catch block may fail with NPE Key: FLINK-8423 URL: https://issues.apache.org/jira/browse/FLINK-8423 Project: Flink Issue Type: Bug Components: Streaming Affects Versions: 1.4.0, 1.5.0 Reporter: Chesnay Schepler Priority: Minor {code} @Override protected void pushToOperator(StreamRecord record) { try { // we know that the given outputTag matches our OutputTag so the record // must be of the type that our operator (and Serializer) expects. @SuppressWarnings("unchecked") StreamRecord castRecord = (StreamRecord) record; numRecordsIn.inc(); StreamRecord copy = castRecord.copy(serializer.copy(castRecord.getValue())); operator.setKeyContextElement1(copy); operator.processElement(copy); } catch (ClassCastException e) { // Enrich error message ClassCastException replace = new ClassCastException( String.format( "%s. Failed to push OutputTag with id '%s' to operator. " + "This can occur when multiple OutputTags with different types " + "but identical names are being used.", e.getMessage(), outputTag.getId())); throw new ExceptionInChainedOperatorException(replace); } catch (Exception e) { throw new ExceptionInChainedOperatorException(e); } } {code} If outputTag is null (as is the case when no sideOutput was defined) the catch block will crash with a NullPointerException. This may happen if {{operator.processElement}} throws a ClassCastException. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Comment Edited] (FLINK-8417) Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer
[ https://issues.apache.org/jira/browse/FLINK-8417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324202#comment-16324202 ] Tzu-Li (Gordon) Tai edited comment on FLINK-8417 at 1/12/18 4:36 PM: - >From the Javadocs, it mentions that the >{{STSAssumeRoleSessionCredentialsProvider}} spawns a background thread that >automatically refreshes the credentials. I assume then that is handled >out-of-the-box by the AWS API? was (Author: tzulitai): Fro the Javadocs, it mentions that the {{STSAssumeRoleSessionCredentialsProvider}} spawns a background thread that automatically refreshes the credentials. I assume then that is handled out-of-the-box by the AWS API? > Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer > --- > > Key: FLINK-8417 > URL: https://issues.apache.org/jira/browse/FLINK-8417 > Project: Flink > Issue Type: New Feature > Components: Kinesis Connector >Reporter: Tzu-Li (Gordon) Tai > Fix For: 1.5.0 > > > As discussed in ML: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Kinesis-Connectors-With-Temporary-Credentials-td17734.html. > Users need the functionality to access cross-account AWS Kinesis streams, > using AWS Temporary Credentials [1]. > We should add support for > {{AWSConfigConstants.CredentialsProvider.STSAssumeRole}}, which internally > would use the {{STSAssumeRoleSessionCredentialsProvider}} [2] in > {{AWSUtil#getCredentialsProvider(Properties)}}. > [1] https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credentials_temp.html > [2] > https://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/com/amazonaws/auth/STSAssumeRoleSessionCredentialsProvider.html -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8417) Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer
[ https://issues.apache.org/jira/browse/FLINK-8417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324204#comment-16324204 ] Tzu-Li (Gordon) Tai commented on FLINK-8417: The Role ARN for the Temporary Credentials should be supplied via the consumer config properties. > Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer > --- > > Key: FLINK-8417 > URL: https://issues.apache.org/jira/browse/FLINK-8417 > Project: Flink > Issue Type: New Feature > Components: Kinesis Connector >Reporter: Tzu-Li (Gordon) Tai > Fix For: 1.5.0 > > > As discussed in ML: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Kinesis-Connectors-With-Temporary-Credentials-td17734.html. > Users need the functionality to access cross-account AWS Kinesis streams, > using AWS Temporary Credentials [1]. > We should add support for > {{AWSConfigConstants.CredentialsProvider.STSAssumeRole}}, which internally > would use the {{STSAssumeRoleSessionCredentialsProvider}} [2] in > {{AWSUtil#getCredentialsProvider(Properties)}}. > [1] https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credentials_temp.html > [2] > https://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/com/amazonaws/auth/STSAssumeRoleSessionCredentialsProvider.html -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8417) Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer
[ https://issues.apache.org/jira/browse/FLINK-8417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324202#comment-16324202 ] Tzu-Li (Gordon) Tai commented on FLINK-8417: Fro the Javadocs, it mentions that the {{STSAssumeRoleSessionCredentialsProvider}} spawns a background thread that automatically refreshes the credentials. I assume then that is handled out-of-the-box by the AWS API? > Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer > --- > > Key: FLINK-8417 > URL: https://issues.apache.org/jira/browse/FLINK-8417 > Project: Flink > Issue Type: New Feature > Components: Kinesis Connector >Reporter: Tzu-Li (Gordon) Tai > Fix For: 1.5.0 > > > As discussed in ML: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Kinesis-Connectors-With-Temporary-Credentials-td17734.html. > Users need the functionality to access cross-account AWS Kinesis streams, > using AWS Temporary Credentials [1]. > We should add support for > {{AWSConfigConstants.CredentialsProvider.STSAssumeRole}}, which internally > would use the {{STSAssumeRoleSessionCredentialsProvider}} [2] in > {{AWSUtil#getCredentialsProvider(Properties)}}. > [1] https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credentials_temp.html > [2] > https://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/com/amazonaws/auth/STSAssumeRoleSessionCredentialsProvider.html -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8422) Checkstyle for org.apache.flink.api.java.tuple
Greg Hogan created FLINK-8422: - Summary: Checkstyle for org.apache.flink.api.java.tuple Key: FLINK-8422 URL: https://issues.apache.org/jira/browse/FLINK-8422 Project: Flink Issue Type: Improvement Components: Core Affects Versions: 1.5.0 Reporter: Greg Hogan Assignee: Greg Hogan Priority: Trivial Update {{TupleGenerator}} for Flink's checkstyle and rebuild {{Tuple}} and {{TupleBuilder}} classes. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Comment Edited] (FLINK-8417) Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer
[ https://issues.apache.org/jira/browse/FLINK-8417?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324096#comment-16324096 ] Sreenath Kodedala edited comment on FLINK-8417 at 1/12/18 4:33 PM: --- STSAssumeRole requires Role ARN(Mandatory) to assume and RoleSessionName (Can set to some default value if not provided) and one thing to consider is these Temporary credentials are set to expire in an hour (Max). How can this be handled without blocking the data flow? was (Author: vedarad): STSAssumeRole requires Role ARN to assume and RoleSessionName (Can set to some default value if not provided) and one thing to consider is these Temporary credentials are set to expire in an hour (Max). How can this be handled without blocking the data flow? > Support STSAssumeRoleSessionCredentialsProvider in FlinkKinesisConsumer > --- > > Key: FLINK-8417 > URL: https://issues.apache.org/jira/browse/FLINK-8417 > Project: Flink > Issue Type: New Feature > Components: Kinesis Connector >Reporter: Tzu-Li (Gordon) Tai > Fix For: 1.5.0 > > > As discussed in ML: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Kinesis-Connectors-With-Temporary-Credentials-td17734.html. > Users need the functionality to access cross-account AWS Kinesis streams, > using AWS Temporary Credentials [1]. > We should add support for > {{AWSConfigConstants.CredentialsProvider.STSAssumeRole}}, which internally > would use the {{STSAssumeRoleSessionCredentialsProvider}} [2] in > {{AWSUtil#getCredentialsProvider(Properties)}}. > [1] https://docs.aws.amazon.com/IAM/latest/UserGuide/id_credentials_temp.html > [2] > https://docs.aws.amazon.com/AWSJavaSDK/latest/javadoc/com/amazonaws/auth/STSAssumeRoleSessionCredentialsProvider.html -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Created] (FLINK-8421) HeapInternalTimerService should reconfigure compatible key / namespace serializers on restore
Tzu-Li (Gordon) Tai created FLINK-8421: -- Summary: HeapInternalTimerService should reconfigure compatible key / namespace serializers on restore Key: FLINK-8421 URL: https://issues.apache.org/jira/browse/FLINK-8421 Project: Flink Issue Type: Bug Affects Versions: 1.4.0, 1.5.0 Reporter: Tzu-Li (Gordon) Tai Priority: Blocker Fix For: 1.3.3, 1.5.0, 1.4.1 The {{HeapInternalTimerService}} still uses simple {{equals}} checks on restored / newly provided serializers for compatibility checks. This should be replaced with the {{TypeSerializer::ensureCompatibility}} checks instead, so that new serializers can be reconfigured. For Flink 1.4.0 release and current master, this is a critical bug since the {{KryoSerializer}} has different default base registrations than before due to FLINK-7420. i.e if the key of a window is serialized using the {{KryoSerializer}} in 1.3.x, the restore would never succeed in 1.4.0. For 1.3.x, this fix would be an improvement, such that the {{HeapInternalTimerService}} restore will make use of serializer reconfiguration. Other remarks: * We need to double check all operators that checkpoint / restore from **raw** state. Apparently, the serializer compatibility checks were only implemented for managed state. * Migration ITCases apparently do not have enough coverage. A migration test job that uses a key type which required the {{KryoSerializer}}, and uses windows, would have caught this issue. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[GitHub] flink pull request #5289: [hotfix] [docs] Fix typos
GitHub user greghogan opened a pull request: https://github.com/apache/flink/pull/5289 [hotfix] [docs] Fix typos From the IntelliJ `Typos` inspection. You can merge this pull request into a Git repository by running: $ git pull https://github.com/greghogan/flink 20180112a_fix_typos Alternatively you can review and apply these changes as the patch at: https://github.com/apache/flink/pull/5289.patch To close this pull request, make a commit to your master/trunk branch with (at least) the following in the commit message: This closes #5289 commit 5658c12c8aafb47654e059bfe7712aa06f78e083 Author: Greg HoganDate: 2018-01-12T14:29:28Z [hotfix] [docs] Fix typos ---
[jira] [Commented] (FLINK-8414) Gelly performance seriously decreases when using the suggested parallelism configuration
[ https://issues.apache.org/jira/browse/FLINK-8414?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324133#comment-16324133 ] Greg Hogan commented on FLINK-8414: --- This is more of a question than a reported bug and may be more appropriate for the flink-user mailing list. Are you able to share what algorithm(s) you are running and describe the dataset(s)? > Gelly performance seriously decreases when using the suggested parallelism > configuration > > > Key: FLINK-8414 > URL: https://issues.apache.org/jira/browse/FLINK-8414 > Project: Flink > Issue Type: Bug > Components: Configuration, Documentation, Gelly >Reporter: flora karniav >Priority: Minor > > I am running Gelly examples with different datasets in a cluster of 5 > machines (1 Jobmanager and 4 Taskmanagers) of 32 cores each. > The number of Slots parameter is set to 32 (as suggested) and the parallelism > to 128 (32 cores*4 taskmanagers). > I observe a vast performance degradation using these suggested settings than > setting parallelism.default to 16 for example were the same job completes at > ~60 seconds vs ~140 in the 128 parallelism case. > Is there something wrong in my configuration? Should I decrease parallelism > and -if so- will this inevitably decrease CPU utilization? > Another matter that may be related to this is the number of partitions of the > data. Is this somehow related to parallelism? How many partitions are created > in the case of parallelism.default=128? -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324117#comment-16324117 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161249788 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); + + /** +* The timeout for a slot request to be discarded, in milliseconds. +*/ + public static final ConfigOption SLOT_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.slot.request-timeout") --- End diff -- maybe `slotmanager.request-timeout` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324113#comment-16324113 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161248207 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions --- End diff -- Let's make this option a `Long`. > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324118#comment-16324118 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161249685 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); + + /** +* The timeout for a slot request to be discarded, in milliseconds. +*/ + public static final ConfigOption SLOT_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.slot.request-timeout") + .defaultValue(60); + + /** +* The timeout for an idle task manager to be released, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.timeout") --- End diff -- maybe `slotmanager.taskmanager-timeout` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324119#comment-16324119 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161249883 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") --- End diff -- maybe `slotmanager.rpc-timeout` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324115#comment-16324115 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161248438 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); + + /** +* The timeout for a slot request to be discarded, in milliseconds. +*/ + public static final ConfigOption SLOT_REQUEST_TIMEOUT = ConfigOptions --- End diff -- `Long` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324114#comment-16324114 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161248457 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); + + /** +* The timeout for a slot request to be discarded, in milliseconds. +*/ + public static final ConfigOption SLOT_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.slot.request-timeout") + .defaultValue(60); + + /** +* The timeout for an idle task manager to be released, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_TIMEOUT = ConfigOptions --- End diff -- `Long` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324116#comment-16324116 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161249010 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); --- End diff -- Please add a description via `withDescription` > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)
[jira] [Commented] (FLINK-8399) Use independent configurations for the different timeouts in slot manager
[ https://issues.apache.org/jira/browse/FLINK-8399?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel=16324120#comment-16324120 ] ASF GitHub Bot commented on FLINK-8399: --- Github user tillrohrmann commented on a diff in the pull request: https://github.com/apache/flink/pull/5271#discussion_r161249992 --- Diff: flink-core/src/main/java/org/apache/flink/configuration/ResourceManagerOptions.java --- @@ -58,6 +58,27 @@ .defaultValue(600) .withDeprecatedKeys("yarn.heap-cutoff-min"); + /** +* The timeout for requesting slot to a task manager, in milliseconds. +*/ + public static final ConfigOption TASK_MANAGER_REQUEST_TIMEOUT = ConfigOptions + .key("slotmanager.taskmanager.request-timeout") + .defaultValue(3); --- End diff -- For the other options, a description would be great as well. > Use independent configurations for the different timeouts in slot manager > - > > Key: FLINK-8399 > URL: https://issues.apache.org/jira/browse/FLINK-8399 > Project: Flink > Issue Type: Bug > Components: Cluster Management >Affects Versions: 1.5.0 >Reporter: shuai.xu >Assignee: shuai.xu > Labels: flip-6 > > There are three parameter in slot manager to indicate the timeout for slot > request to task manager, slot request to be discarded and task manager to be > released. But now they all come from the value of AkkaOptions.ASK_TIMEOUT, > need to use independent configurations for them. -- This message was sent by Atlassian JIRA (v6.4.14#64029)