Re: Apache Tinkerpop & Geode Integration?
Hi James, I've just subscribed to the Tinkerpop dev mailing list. Could you please send a reply to the thread, so then I can reply to it? I'm not sure how I can reply to the thread otherwise... I also saw that there is a grafos.ml project thread. I could also provide some input there :) Thanks! -Vasia. On 25 November 2015 at 15:09, James Thorntonwrote: > Hi Vasia - > > Yes, a FlinkGraphComputer should be a straight-forward first step. Also, on > the Apache Tinkerpop dev mailing list, Marko thought it might be cool if > there was a "Graph API" similar to the "Table API" -- hooking in Gremlin to > Flink's fluent API would give Flink users a full graph query language. > > Stephen Mallette is a TinkerPop core contributor, and he has already > started working on a FlinkGraphComputer. There is a Flink/Tinkerpop thread > on the TinkerPop dev list -- it would be great to have you part of the > conversation there too as we work on the integration: > >http://mail-archives.apache.org/mod_mbox/incubator-tinkerpop-dev/ > > Thanks, Vasia. > > - James > > > On Mon, Nov 23, 2015 at 10:28 AM, Vasiliki Kalavri < > vasilikikala...@gmail.com> wrote: > > > Hi James, > > > > thank you for your e-mail and your interest in Flink :) > > > > I've recently taken a _quick_ look into Apache TinkerPop and I think it'd > > be very interesting to integrate with Flink/Gelly. > > Are you thinking about something like a Flink GraphComputer, similar to > > Giraph and Spark GraphComputer's? > > I believe such an integration should be straight-forward to implement. > You > > can start by looking into Flink iteration operators [1] and Gelly > iteration > > abstractions [2]. > > > > Regarding Apache Geode, I'm not familiar with project, but I'll try to > take > > a look in the following days! > > > > Cheers, > > -Vasia. > > > > > > [1]: > > > > > https://ci.apache.org/projects/flink/flink-docs-master/apis/programming_guide.html#iteration-operators > > [2]: > > > > > https://ci.apache.org/projects/flink/flink-docs-master/libs/gelly_guide.html#iterative-graph-processing > > > > > > On 20 November 2015 at 08:32, James Thornton > > wrote: > > > > > Hi - > > > > > > This is James Thornton (espeed) from the Apache Tinkerpop project ( > > > http://tinkerpop.incubator.apache.org/). > > > > > > The Flink iterators should pair well with Gremlin's Graph Traversal > > Machine > > > ( > > > > > > > > > http://www.datastax.com/dev/blog/the-benefits-of-the-gremlin-graph-traversal-machine > > > ) > > > -- it would be good to coordinate on creating an integration. > > > > > > Also, Apache Geode made a splash today on HN ( > > > https://news.ycombinator.com/item?id=10596859) -- connecting Geode and > > > Flink would be killer. Here's the Geode/Spark connector for > refefference: > > > > > > > > > > > > > > > https://github.com/apache/incubator-geode/tree/develop/gemfire-spark-connector > > > > > > - James > > > > > >
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
+1 I ran an example with a custom operator that processes high-volume kafka input/output and has a large state size. I ran this on 10 GCE nodes. > On 25 Nov 2015, at 14:58, Till Rohrmannwrote: > > Alright, then I withdraw my remark concerning testdata.avro. > > On Wed, Nov 25, 2015 at 2:56 PM, Stephan Ewen wrote: > >> @Till I think the avro test data file is okay, the "no binaries" policy >> refers to binary executables, as far as I know. >> >> On Wed, Nov 25, 2015 at 2:54 PM, Till Rohrmann >> wrote: >> >>> Checked checksums for src release and Hadoop 2.7 Scala 2.10 release >>> >>> Checked binaries in source release >>> - contains ./flink-staging/flink-avro/src/test/resources/testdata.avro >>> >>> License >>> - no new files added which are relevant for licensing >>> >>> Build Flink and run tests from source release for Hadoop 2.5.1 >>> >>> Checked empty that log files don't contain exceptions and out files are >>> empty >>> >>> Run all examples with Hadoop 2.7 Scala 2.10 binaries via FliRTT tool on 4 >>> node standalone cluster and YARN cluster >>> >>> Tested planVisualizer >>> >>> Tested flink command line client >>> - tested info command >>> - tested -p option >>> >>> Tested cluster HA in standalone mode => working >>> >>> Tested cluster HA on YARN (2.7.1) => working >>> >>> Except for the avro testdata file which is contained in the source >> release, >>> I didn't find anything. >>> >>> +1 for releasing and removing the testdata file for the next release. >>> >>> On Wed, Nov 25, 2015 at 2:33 PM, Robert Metzger >>> wrote: >>> +1 - Build a maven project with the staging repository - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster >>> with YARN and HDFS HA - ran some kafka (0.8.2.0) read / write experiments - job cancellation with yarn is working ;) I found the following issue while testing: https://issues.apache.org/jira/browse/FLINK-3078 but it was already in 0.10.0 and its not super critical bc the JobManager container will be killed by YARN after a few minutes. I'll extend the vote until tomorrow Thursday, November 26. On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewen >> wrote: > @Gyula: I think it affects users, so should definitely be fixed very >>> soon > (either 0.10.1 or 0.10.2) > > Still checking whether Robert's current version fix solves it now, or > not... > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > vyacheslav.zholu...@gmail.com> wrote: > >> I can confirm that the build works fine when increasing a max >> number >>> of >> opened files. Sorry for confusion. >> >> >> >> -- >> View this message in context: >> > >>> >> http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html >> Sent from the Apache Flink Mailing List archive. mailing list >> archive at >> Nabble.com. >> > >>> >>
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
+1 - Verified hashes and signatures - Ran example jobs on YARN with vanilla Hadoop vesions (on 4 GCE nodes): * 2.7.1 with Flink Hadoop 2.7 binary, Scala 2.10 and 11 * 2.6.2 with Flink Hadoop 2.6 binary, Scala 2.10 * 2.4.1 with Flink Hadoop 2.4 binary, Scala 2.10 * 2.3.0 with Flink Hadoop 2 binary, Scala 2.10 - Cancelled a restarting job via CLI and web interface - Ran simple Kafka read-write pipeline - Ran manual tests - Ran examples on local cluster > On 25 Nov 2015, at 17:45, Aljoscha Krettekwrote: > > +1 > > I ran an example with a custom operator that processes high-volume kafka > input/output and has a large state size. I ran this on 10 GCE nodes. > >> On 25 Nov 2015, at 14:58, Till Rohrmann wrote: >> >> Alright, then I withdraw my remark concerning testdata.avro. >> >> On Wed, Nov 25, 2015 at 2:56 PM, Stephan Ewen wrote: >> >>> @Till I think the avro test data file is okay, the "no binaries" policy >>> refers to binary executables, as far as I know. >>> >>> On Wed, Nov 25, 2015 at 2:54 PM, Till Rohrmann >>> wrote: >>> Checked checksums for src release and Hadoop 2.7 Scala 2.10 release Checked binaries in source release - contains ./flink-staging/flink-avro/src/test/resources/testdata.avro License - no new files added which are relevant for licensing Build Flink and run tests from source release for Hadoop 2.5.1 Checked empty that log files don't contain exceptions and out files are empty Run all examples with Hadoop 2.7 Scala 2.10 binaries via FliRTT tool on 4 node standalone cluster and YARN cluster Tested planVisualizer Tested flink command line client - tested info command - tested -p option Tested cluster HA in standalone mode => working Tested cluster HA on YARN (2.7.1) => working Except for the avro testdata file which is contained in the source >>> release, I didn't find anything. +1 for releasing and removing the testdata file for the next release. On Wed, Nov 25, 2015 at 2:33 PM, Robert Metzger wrote: > +1 > > - Build a maven project with the staging repository > - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster with > YARN and HDFS HA > - ran some kafka (0.8.2.0) read / write experiments > - job cancellation with yarn is working ;) > > I found the following issue while testing: > https://issues.apache.org/jira/browse/FLINK-3078 but it was already in > 0.10.0 and its not super critical bc the JobManager container will be > killed by YARN after a few minutes. > > > I'll extend the vote until tomorrow Thursday, November 26. > > > On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewen >>> wrote: > >> @Gyula: I think it affects users, so should definitely be fixed very soon >> (either 0.10.1 or 0.10.2) >> >> Still checking whether Robert's current version fix solves it now, or >> not... >> >> On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < >> vyacheslav.zholu...@gmail.com> wrote: >> >>> I can confirm that the build works fine when increasing a max >>> number of >>> opened files. Sorry for confusion. >>> >>> >>> >>> -- >>> View this message in context: >>> >> > >>> http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html >>> Sent from the Apache Flink Mailing List archive. mailing list >>> archive > at >>> Nabble.com. >>> >> > >>> >
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
+1 LICENSE file looks good in source artifact NOTICE file looks good in source artifact Signature file looks good in source artifact Hash files looks good in source artifact No 3rd party executables in source artifact Source compiled All tests are passed Run standalone mode test app - Henry On Mon, Nov 23, 2015 at 4:45 AM, Robert Metzgerwrote: > Hi All, > > this is the first bugfix release for the 0.10 series of Flink. > I've CC'ed the user@ list if users are interested in helping to verify the > release. > > It contains fixes for critical issues, in particular: > - FLINK-3021 Fix class loading issue for streaming sources > - FLINK-2974 Add periodic offset committer for Kafka > - FLINK-2977 Using reflection to load HBase Kerberos tokens > - FLINK-3024 Fix TimestampExtractor.getCurrentWatermark() Behaviour > - FLINK-2967 Increase timeout for LOCAL_HOST address detection stratey > - FLINK-3025 [kafka consumer] Bump transitive ZkClient dependency > - FLINK-2989 job cancel button doesn't work on YARN > - FLINK-3032: Flink does not start on Hadoop 2.7.1 (HDP), due to class > conflict > - FLINK-3011, 3019, 3028 Cancel jobs in RESTARTING state > > This is the guide on how to verify a release: > https://cwiki.apache.org/confluence/display/FLINK/Releasing > > During the testing, please focus on trying out Flink on different Hadoop > platforms: We changed the way how Hadoop's Maven dependencies are packaged, > so maybe there are issues with different Hadoop distributions. > The Kafka consumer also changed a bit, would be good to test it on a > cluster. > > - > > Please vote on releasing the following candidate as Apache Flink version > 0.10.1: > > The commit to be voted on: > http://git-wip-us.apache.org/repos/asf/flink/commit/2e9b2316 > > Branch: > release-0.10.1-rc1 (see > https://git1-us-west.apache.org/repos/asf/flink/?p=flink.git) > > The release artifacts to be voted on can be found at: > http://people.apache.org/~rmetzger/flink-0.10.1-rc1/ > > The release artifacts are signed with the key with fingerprint D9839159: > http://www.apache.org/dist/flink/KEYS > > The staging repository for this release can be found at: > https://repository.apache.org/content/repositories/orgapacheflink-1058 > > - > > The vote is open for the next 72 hours and passes if a majority of at least > three +1 PMC votes are cast. > > The vote ends on Wednesday, November 25. > > [ ] +1 Release this package as Apache Flink 0.10.1 > [ ] -1 Do not release this package because ... > > ===
[jira] [Created] (FLINK-3079) Add utility for measuring the raw read throughput from a Kafka topic
Robert Metzger created FLINK-3079: - Summary: Add utility for measuring the raw read throughput from a Kafka topic Key: FLINK-3079 URL: https://issues.apache.org/jira/browse/FLINK-3079 Project: Flink Issue Type: New Feature Components: Kafka Connector Reporter: Robert Metzger Assignee: Robert Metzger Priority: Minor I would like to add a little example job to the kafka connector package which is measuring the raw read throughput from a kafka topic. This is helpful when trying to debug the performance when reading from Kafka. Such a tool would help finding the bottleneck in a data processing pipeline. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (FLINK-3080) Cannot union a data stream with a product of itself
Vasia Kalavri created FLINK-3080: Summary: Cannot union a data stream with a product of itself Key: FLINK-3080 URL: https://issues.apache.org/jira/browse/FLINK-3080 Project: Flink Issue Type: Bug Components: Streaming Affects Versions: 0.10.0 Reporter: Vasia Kalavri Currently it is not possible to union a stream with itself or a product of itself. For example, the following code {{stream.union(stream.map(...))}} fails with this exception: "A DataStream cannot be unioned with itself". The documentation currently states that "If you union a data stream with itself you will still only get each element once." and should also be updated. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: Union a data stream with a product of itself
Here's the issue: https://issues.apache.org/jira/browse/FLINK-3080 -V. On 25 November 2015 at 14:38, Gyula Fórawrote: > Yes, please > > Vasiliki Kalavri ezt írta (időpont: 2015. nov. > 25., Sze, 14:37): > > > So, do we all agree that the current behavior is not correct? Shall I > open > > a JIRA about this? > > > > On 25 November 2015 at 13:58, Gyula Fóra wrote: > > > > > Well it kind of depends on what definition of union are we using. If > this > > > is a union in a set theoretical way we can argue that the union of a > > stream > > > with itself should be the same stream because it contains exactly the > > same > > > elements with the same timestamps and lineage. > > > > > > On the other hand stream and stream.map(id) are not exactly the same as > > > they might have elements with different order (the lineage differs). > > > > > > So I wouldnt say that any self-union semantics is the only possible > one. > > > > > > Gyula > > > > > > Bruecke, Christoph ezt írta > > > (időpont: 2015. nov. 25., Sze, 13:47): > > > > > > > Hi, > > > > > > > > the operation “stream.union(stream.map(id))” is equivalent to > > > > “stream.union(stream)” isn’t it? So it might also duplicate the data. > > > > > > > > - Christoph > > > > > > > > > > > > > On 25 Nov 2015, at 11:24, Stephan Ewen wrote: > > > > > > > > > > "stream.union(stream.map(..))" should definitely be possible. Not > > sure > > > > why > > > > > this is not permitted. > > > > > > > > > > "stream.union(stream)" would contain each element twice, so should > > > either > > > > > give an error or actually union (or duplicate) elements... > > > > > > > > > > Stephan > > > > > > > > > > > > > > > On Wed, Nov 25, 2015 at 10:42 AM, Gyula Fóra > > > wrote: > > > > > > > > > >> Yes, I am not sure if this the intentional behaviour. I think you > > are > > > > >> supposed to be able to do the things you described. > > > > >> > > > > >> stream.union(stream.map(..)) and things like this are fair > > operations. > > > > Also > > > > >> maybe stream.union(stream) should just give stream instead of an > > > error. > > > > >> > > > > >> Could someone comment on this who knows the reasoning behind the > > > current > > > > >> mechanics? > > > > >> > > > > >> Gyula > > > > >> > > > > >> Vasiliki Kalavri ezt írta (időpont: > > 2015. > > > > nov. > > > > >> 24., K, 16:46): > > > > >> > > > > >>> Hi squirrels, > > > > >>> > > > > >>> when porting the gelly streaming code from 0.9 to 0.10 today with > > > > Paris, > > > > >> we > > > > >>> hit an exception in union: "*A DataStream cannot be unioned with > > > > >> itself*". > > > > >>> > > > > >>> The code raising this exception looks like this: > > > > >>> stream.union(stream.map(...)). > > > > >>> > > > > >>> Taking a look into the union code, we see that it's now not > allowed > > > to > > > > >>> union a stream, not only with itself, but with any product of > > itself. > > > > >>> > > > > >>> First, we are wondering, why is that? Does it make building the > > > stream > > > > >>> graph easier in some way? > > > > >>> Second, we might want to give a better error message there, e.g. > > "*A > > > > >>> DataStream cannot be unioned with itself or a product of > itself*", > > > and > > > > >>> finally, we should update the docs, which currently state that > > union > > > a > > > > >>> stream with itself is allowed and that "*If you union a data > stream > > > > with > > > > >>> itself you will still only get each element once.*" > > > > >>> > > > > >>> Cheers, > > > > >>> -Vasia. > > > > >>> > > > > >> > > > > > > > > > > > > > >
Re: Null Pointer Exception in tests but only in COLLECTION mode
Hi Martin, Great. Thanks for the fix! Cheers, Max On Tue, Nov 24, 2015 at 7:40 PM, Martin Junghannswrote: > Hi Max, > > fixed in https://github.com/apache/flink/pull/1396 > > Best, > Martin > > > On 24.11.2015 13:46, Maximilian Michels wrote: >> >> Hi André, hi Martin, >> >> This looks very much like a bug. Martin, I would be happy if you >> opened a JIRA issue. >> >> Thanks, >> Max >> >> On Sun, Nov 22, 2015 at 12:27 PM, Martin Junghanns >> wrote: >>> >>> Hi, >>> >>> What he meant was MultipleProgramsTestBase, not FlinkTestBase. >>> >>> I debugged this a bit. >>> >>> The NPE is thrown in >>> >>> >>> https://github.com/apache/flink/blob/master/flink-java/src/main/java/org/apache/flink/api/java/operators/AggregateOperator.java#L296 >>> >>> since current can be null if the input iterator is empty. >>> >>> In Cluster Execution, it is checked that the output of the previous >>> function >>> (e.g. Filter) is not empty in: >>> >>> >>> https://github.com/apache/flink/blob/master/flink-runtime/src/main/java/org/apache/flink/runtime/operators/AllGroupReduceDriver.java#L144 >>> >>> which avoids going into AggregateOperator and getting a NPE. >>> >>> However, in Collection Mode, the execution is not grouped (don't know >>> why, >>> yet). In >>> >>> >>> https://github.com/apache/flink/blob/master/flink-core/src/main/java/org/apache/flink/api/common/operators/base/GroupReduceOperatorBase.java#L207 >>> >>> the copied input data is handed over to the aggregate function which >>> leads >>> to the NPE. >>> >>> Checking inputDataCopy.size() > 0 before calling the aggregate solves the >>> problem. >>> >>> If someone can confirm that this is not a more generic problem, I would >>> open >>> an issue and a PR. >>> >>> Best, >>> Martin >>> >>> >>> On 20.11.2015 18:41, André Petermann wrote: Hi all, during a workflow, a data set may run empty, e.g., because of a join without matches. We're using FlinkTestBase and found out, that aggregate functions on empty data sets work fine in CLUSTER execution mode but cause a Null Pointer Exception at AggregateOperator$AggregatingUdf in COLLECTION mode. Here is the minimal example on 1.0-SNAPSHOT: https://gist.github.com/p3et/59a65bab11098dd11054 Are we doing something wrong, or is this a bug? Cheers, Andre >>> >
RE: The null in Flink
Hi In this mail list, there are some discussions about null value handling in Flink, and I saw several related JIRAs as well(like FLINK-2203, FLINK-2210), but unfortunately, got reverted due to immature design, and no further action since then. I would like to pick this topic up here, as it's quite an important part of data analysis and many features depend on it. Hopefully, through a plenary discussion, we can generate an acceptable solution and move forward. Stephan has explained very clearly about how and why Flink handle "Null values in the Programming Language APIs", so I mainly talk about the second part of "Null values in the high-level (logical) APIs ". 1. Why should Flink support Null values handling in Table API? i. Data source may miss column value in many cases, if no Null values handling in Table API, user need to write an extra ETL to handle missing values manually. ii. Some Table API operators generate Null values on their own, like Outer Join/Cube/Rollup/Grouping Set, and so on. Null values handling in Table API is the prerequisite of these features. 2. The semantic of Null value handling in Table API. Fortunately, there are already mature DBMS standards we can follow for Null value handling, I list several semantic of Null value handling here. To be noted that, this may not cover all the cases, and the semantics may vary in different DBMSs, so it should totally open to discuss. I, NULL compare. In ascending order, NULL is smaller than any other value, and NULL == NULL return false. ii. NULL exists in GroupBy Key, all NULL values are grouped as a single group. iii. NULL exists in Aggregate columns, ignore NULL in aggregation function. iv. NULL exists in both side Join key, refer to #i, NULL == NULL return false, no output for NULL Join key. v. NULL in Scalar expression, expression within NULL(eg. 1 + NULL) return NULL. vi. NULL in Boolean expression, add an extra result: UNKNOWN, more semantic for Boolean expression in reference #1. vii. More related function support, like COALESCE, NVL, NANVL, and so on. 3. NULL value storage in Table API. Just set null to Row field value. To mark NULL value in serialized binary record data, normally it use extra flag for each field to mark whether its value is NULL, which would change the data layout of Row object. So any logic that access serialized Row data directly should updated to sync with new data layout, for example, many methods in RowComparator. Reference: 1. Nulls: Nothing to worry about: http://www.oracle.com/technetwork/issue-archive/2005/05-jul/o45sql-097727.html. 2. Null related functions: https://oracle-base.com/articles/misc/null-related-functions -Original Message- From: ewenstep...@gmail.com [mailto:ewenstep...@gmail.com] On Behalf Of Stephan Ewen Sent: Thursday, June 18, 2015 8:43 AM To: dev@flink.apache.org Subject: Re: The null in Flink Hi! I think we actually have two discussions here, both of them important: -- 1) Null values in the Programming Language APIs -- Fields in composite types may simply be null pointers. In object types: - primitives members are naturally non-nullable - all other members are nullable => If you want to avoid the overhead of nullability, go with primitive types. In Tuples, and derives types (Scala case classes): - Fields are non-nullable. => The reason here is that we initially decided to keep tuples as a very fast data type. Because tuples cannot hold primitives in Java/Scala, we would not have a way to make fast non-nullable fields. The performance of nullable fields affects the key-operations, especially on normalized keys. We can work around that with some effort, but have not one it so far. => In Scala, the Option types is a natural way of elegantly working around that. -- 2) Null values in the high-level (logial) APIs -- This is mainly what Ted was referring to, if I understood him correctly. Here, we need to figure out what form of semantical null values in the Table API and later, in SQL. Besides deciding what semantics to follow here in the logical APIs, we need to decide what these values confert to/from when switching between logical/physical APIs. On Mon, Jun 15, 2015 at 10:07 AM, Ted Dunningwrote: > On Mon, Jun 15, 2015 at 8:45 AM, Maximilian Michels > wrote: > > > Just to give an idea what null values could cause in Flink: > DataSet.count() > > returns the number of elements of all values in a Dataset (null or > > not) while #834 would ignore null values and aggregate the DataSet > > without > them. > > > > Compare R's
Re: could you please add me Contributor list?
I added you as a contributor. On Wed, Nov 25, 2015 at 7:29 AM, jun aokiwrote: > Hi Henry, thank you for helping. My id is jaoki > > On Tue, Nov 24, 2015 at 9:47 PM, Henry Saputra > wrote: > > > Hi Jun, > > > > What is your JIRA username? > > > > But for now, you can always work on JIRA issue without assigning to > > yourself. Just add to comment that you are planning to work on it. > > > > - Henry > > > > On Tue, Nov 24, 2015 at 5:18 PM, jun aoki wrote: > > > So that I can assign myself to Flink jiras? > > > > > > -- > > > -jun > > > > > > -- > -jun >
Re: Union a data stream with a product of itself
Yes, I am not sure if this the intentional behaviour. I think you are supposed to be able to do the things you described. stream.union(stream.map(..)) and things like this are fair operations. Also maybe stream.union(stream) should just give stream instead of an error. Could someone comment on this who knows the reasoning behind the current mechanics? Gyula Vasiliki Kalavriezt írta (időpont: 2015. nov. 24., K, 16:46): > Hi squirrels, > > when porting the gelly streaming code from 0.9 to 0.10 today with Paris, we > hit an exception in union: "*A DataStream cannot be unioned with itself*". > > The code raising this exception looks like this: > stream.union(stream.map(...)). > > Taking a look into the union code, we see that it's now not allowed to > union a stream, not only with itself, but with any product of itself. > > First, we are wondering, why is that? Does it make building the stream > graph easier in some way? > Second, we might want to give a better error message there, e.g. "*A > DataStream cannot be unioned with itself or a product of itself*", and > finally, we should update the docs, which currently state that union a > stream with itself is allowed and that "*If you union a data stream with > itself you will still only get each element once.*" > > Cheers, > -Vasia. >
Re: Union a data stream with a product of itself
"stream.union(stream.map(..))" should definitely be possible. Not sure why this is not permitted. "stream.union(stream)" would contain each element twice, so should either give an error or actually union (or duplicate) elements... Stephan On Wed, Nov 25, 2015 at 10:42 AM, Gyula Fórawrote: > Yes, I am not sure if this the intentional behaviour. I think you are > supposed to be able to do the things you described. > > stream.union(stream.map(..)) and things like this are fair operations. Also > maybe stream.union(stream) should just give stream instead of an error. > > Could someone comment on this who knows the reasoning behind the current > mechanics? > > Gyula > > Vasiliki Kalavri ezt írta (időpont: 2015. nov. > 24., K, 16:46): > > > Hi squirrels, > > > > when porting the gelly streaming code from 0.9 to 0.10 today with Paris, > we > > hit an exception in union: "*A DataStream cannot be unioned with > itself*". > > > > The code raising this exception looks like this: > > stream.union(stream.map(...)). > > > > Taking a look into the union code, we see that it's now not allowed to > > union a stream, not only with itself, but with any product of itself. > > > > First, we are wondering, why is that? Does it make building the stream > > graph easier in some way? > > Second, we might want to give a better error message there, e.g. "*A > > DataStream cannot be unioned with itself or a product of itself*", and > > finally, we should update the docs, which currently state that union a > > stream with itself is allowed and that "*If you union a data stream with > > itself you will still only get each element once.*" > > > > Cheers, > > -Vasia. > > >
Re: The null in Flink
Hi Chengxiang, I totally agree that the Table API should fully support NULL values. The Table API is a logical API and therefore we should be as close to ANSI SQL as possible. Rows need to be nullable in the near future. 2. i, ii, iii and iv sound reasonable. But v, vi and vii sound to much like SQL magic. I think all other SQL magic (DBMS specific corner cases) should be handled by the SQL API on top of the Table API. Regards, Timo On 25.11.2015 11:31, Li, Chengxiang wrote: Hi In this mail list, there are some discussions about null value handling in Flink, and I saw several related JIRAs as well(like FLINK-2203, FLINK-2210), but unfortunately, got reverted due to immature design, and no further action since then. I would like to pick this topic up here, as it's quite an important part of data analysis and many features depend on it. Hopefully, through a plenary discussion, we can generate an acceptable solution and move forward. Stephan has explained very clearly about how and why Flink handle "Null values in the Programming Language APIs", so I mainly talk about the second part of "Null values in the high-level (logical) APIs ". 1. Why should Flink support Null values handling in Table API? i. Data source may miss column value in many cases, if no Null values handling in Table API, user need to write an extra ETL to handle missing values manually. ii. Some Table API operators generate Null values on their own, like Outer Join/Cube/Rollup/Grouping Set, and so on. Null values handling in Table API is the prerequisite of these features. 2. The semantic of Null value handling in Table API. Fortunately, there are already mature DBMS standards we can follow for Null value handling, I list several semantic of Null value handling here. To be noted that, this may not cover all the cases, and the semantics may vary in different DBMSs, so it should totally open to discuss. I, NULL compare. In ascending order, NULL is smaller than any other value, and NULL == NULL return false. ii. NULL exists in GroupBy Key, all NULL values are grouped as a single group. iii. NULL exists in Aggregate columns, ignore NULL in aggregation function. iv. NULL exists in both side Join key, refer to #i, NULL == NULL return false, no output for NULL Join key. v. NULL in Scalar expression, expression within NULL(eg. 1 + NULL) return NULL. vi. NULL in Boolean expression, add an extra result: UNKNOWN, more semantic for Boolean expression in reference #1. vii. More related function support, like COALESCE, NVL, NANVL, and so on. 3. NULL value storage in Table API. Just set null to Row field value. To mark NULL value in serialized binary record data, normally it use extra flag for each field to mark whether its value is NULL, which would change the data layout of Row object. So any logic that access serialized Row data directly should updated to sync with new data layout, for example, many methods in RowComparator. Reference: 1. Nulls: Nothing to worry about: http://www.oracle.com/technetwork/issue-archive/2005/05-jul/o45sql-097727.html. 2. Null related functions: https://oracle-base.com/articles/misc/null-related-functions -Original Message- From: ewenstep...@gmail.com [mailto:ewenstep...@gmail.com] On Behalf Of Stephan Ewen Sent: Thursday, June 18, 2015 8:43 AM To: dev@flink.apache.org Subject: Re: The null in Flink Hi! I think we actually have two discussions here, both of them important: -- 1) Null values in the Programming Language APIs -- Fields in composite types may simply be null pointers. In object types: - primitives members are naturally non-nullable - all other members are nullable => If you want to avoid the overhead of nullability, go with primitive types. In Tuples, and derives types (Scala case classes): - Fields are non-nullable. => The reason here is that we initially decided to keep tuples as a very fast data type. Because tuples cannot hold primitives in Java/Scala, we would not have a way to make fast non-nullable fields. The performance of nullable fields affects the key-operations, especially on normalized keys. We can work around that with some effort, but have not one it so far. => In Scala, the Option types is a natural way of elegantly working around that. -- 2) Null values in the high-level (logial) APIs -- This is mainly what Ted was referring to, if I understood him correctly. Here, we need to figure out what form of semantical null values in the Table API and later, in SQL. Besides deciding what semantics to follow here in the logical APIs, we need to decide what these values confert
[jira] [Created] (FLINK-3077) Add "version" command to CliFrontend for showing the version of the installation
Robert Metzger created FLINK-3077: - Summary: Add "version" command to CliFrontend for showing the version of the installation Key: FLINK-3077 URL: https://issues.apache.org/jira/browse/FLINK-3077 Project: Flink Issue Type: Improvement Components: Command-line client Reporter: Robert Metzger Fix For: 1.0.0 I have the bin directory of Flink in my $PATH variable, so I can just do "flink run" on the command line for executing stuff. However, I have multiple Flink versions locally and its hard to find out which installation the bash is picking in the end. adding a simple "version" command will resolve that issue and I consider it helpful in general. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
[jira] [Created] (FLINK-3078) JobManager does not shutdown when checkpointed jobs are running
Robert Metzger created FLINK-3078: - Summary: JobManager does not shutdown when checkpointed jobs are running Key: FLINK-3078 URL: https://issues.apache.org/jira/browse/FLINK-3078 Project: Flink Issue Type: Bug Components: JobManager Affects Versions: 0.10.0, 0.10.1 Reporter: Robert Metzger While testing the 0.10.1 release, I found that the JobManager does not shutdown when I'm stopping it while a streaming job is running. It seems that the checkpoint coordinator and the execution graph are still logging, even though the JobManager actor system and other services have been shut down. This is a log file of an affected JobManager: https://gist.github.com/rmetzger/a1532c18eb7081977cee {code} 11:58:04,406 INFO org.apache.flink.runtime.executiongraph.ExecutionGraph - Flat Map -> Sink: Unnamed (10/10) (c6544ca6d88e2d1acdec5c838d5fce06) switched from CANCELING to FAILED 11:58:04,406 DEBUG org.apache.flink.runtime.executiongraph.ExecutionGraph - Kafka Consumer Topology switched from FAILING to RESTARTING. 11:58:04,407 INFO org.apache.flink.runtime.executiongraph.ExecutionGraph - Delaying retry of job execution for 10 ms ... 11:58:04,417 INFO org.apache.flink.runtime.blob.BlobServer - Stopped BLOB server at 0.0.0.0:44904 11:58:04,421 INFO akka.remote.RemoteActorRefProvider$RemotingTerminator - Shutting down remote daemon. 11:58:04,422 INFO akka.remote.RemoteActorRefProvider$RemotingTerminator - Remote daemon shut down; proceeding with flushing remote transports. 11:58:04,446 INFO akka.remote.RemoteActorRefProvider$RemotingTerminator - Remoting shut down. 11:58:04,473 INFO org.apache.flink.runtime.webmonitor.WebRuntimeMonitor - Removing web root dir /tmp/flink-web-2039bed3-d9f9-4950-83ab-6fb70f7fc302 11:58:04,590 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Triggering checkpoint 66 @ 1448452684590 11:58:04,590 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Checkpoint triggering task Source: Custom Source (1/10) is not being executed at the moment. Aborting checkpoint. 11:58:05,091 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Triggering checkpoint 67 @ 1448452685091 11:58:05,091 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Checkpoint triggering task Source: Custom Source (1/10) is not being executed at the moment. Aborting checkpoint. 11:58:05,590 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Triggering checkpoint 68 @ 1448452685590 11:58:05,590 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Checkpoint triggering task Source: Custom Source (1/10) is not being executed at the moment. Aborting checkpoint. 11:58:06,090 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Triggering checkpoint 69 @ 1448452686090 11:58:06,091 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Checkpoint triggering task Source: Custom Source (1/10) is not being executed at the moment. Aborting checkpoint. 11:58:06,590 INFO org.apache.flink.runtime.checkpoint.CheckpointCoordinator - Triggering checkpoint 70 @ 1448452686590 {code} -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
+1 - Build a maven project with the staging repository - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster with YARN and HDFS HA - ran some kafka (0.8.2.0) read / write experiments - job cancellation with yarn is working ;) I found the following issue while testing: https://issues.apache.org/jira/browse/FLINK-3078 but it was already in 0.10.0 and its not super critical bc the JobManager container will be killed by YARN after a few minutes. I'll extend the vote until tomorrow Thursday, November 26. On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewenwrote: > @Gyula: I think it affects users, so should definitely be fixed very soon > (either 0.10.1 or 0.10.2) > > Still checking whether Robert's current version fix solves it now, or > not... > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > vyacheslav.zholu...@gmail.com> wrote: > > > I can confirm that the build works fine when increasing a max number of > > opened files. Sorry for confusion. > > > > > > > > -- > > View this message in context: > > > http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html > > Sent from the Apache Flink Mailing List archive. mailing list archive at > > Nabble.com. > > >
Re: Union a data stream with a product of itself
Well it kind of depends on what definition of union are we using. If this is a union in a set theoretical way we can argue that the union of a stream with itself should be the same stream because it contains exactly the same elements with the same timestamps and lineage. On the other hand stream and stream.map(id) are not exactly the same as they might have elements with different order (the lineage differs). So I wouldnt say that any self-union semantics is the only possible one. Gyula Bruecke, Christophezt írta (időpont: 2015. nov. 25., Sze, 13:47): > Hi, > > the operation “stream.union(stream.map(id))” is equivalent to > “stream.union(stream)” isn’t it? So it might also duplicate the data. > > - Christoph > > > > On 25 Nov 2015, at 11:24, Stephan Ewen wrote: > > > > "stream.union(stream.map(..))" should definitely be possible. Not sure > why > > this is not permitted. > > > > "stream.union(stream)" would contain each element twice, so should either > > give an error or actually union (or duplicate) elements... > > > > Stephan > > > > > > On Wed, Nov 25, 2015 at 10:42 AM, Gyula Fóra wrote: > > > >> Yes, I am not sure if this the intentional behaviour. I think you are > >> supposed to be able to do the things you described. > >> > >> stream.union(stream.map(..)) and things like this are fair operations. > Also > >> maybe stream.union(stream) should just give stream instead of an error. > >> > >> Could someone comment on this who knows the reasoning behind the current > >> mechanics? > >> > >> Gyula > >> > >> Vasiliki Kalavri ezt írta (időpont: 2015. > nov. > >> 24., K, 16:46): > >> > >>> Hi squirrels, > >>> > >>> when porting the gelly streaming code from 0.9 to 0.10 today with > Paris, > >> we > >>> hit an exception in union: "*A DataStream cannot be unioned with > >> itself*". > >>> > >>> The code raising this exception looks like this: > >>> stream.union(stream.map(...)). > >>> > >>> Taking a look into the union code, we see that it's now not allowed to > >>> union a stream, not only with itself, but with any product of itself. > >>> > >>> First, we are wondering, why is that? Does it make building the stream > >>> graph easier in some way? > >>> Second, we might want to give a better error message there, e.g. "*A > >>> DataStream cannot be unioned with itself or a product of itself*", and > >>> finally, we should update the docs, which currently state that union a > >>> stream with itself is allowed and that "*If you union a data stream > with > >>> itself you will still only get each element once.*" > >>> > >>> Cheers, > >>> -Vasia. > >>> > >> > >
Re: Union a data stream with a product of itself
So, do we all agree that the current behavior is not correct? Shall I open a JIRA about this? On 25 November 2015 at 13:58, Gyula Fórawrote: > Well it kind of depends on what definition of union are we using. If this > is a union in a set theoretical way we can argue that the union of a stream > with itself should be the same stream because it contains exactly the same > elements with the same timestamps and lineage. > > On the other hand stream and stream.map(id) are not exactly the same as > they might have elements with different order (the lineage differs). > > So I wouldnt say that any self-union semantics is the only possible one. > > Gyula > > Bruecke, Christoph ezt írta > (időpont: 2015. nov. 25., Sze, 13:47): > > > Hi, > > > > the operation “stream.union(stream.map(id))” is equivalent to > > “stream.union(stream)” isn’t it? So it might also duplicate the data. > > > > - Christoph > > > > > > > On 25 Nov 2015, at 11:24, Stephan Ewen wrote: > > > > > > "stream.union(stream.map(..))" should definitely be possible. Not sure > > why > > > this is not permitted. > > > > > > "stream.union(stream)" would contain each element twice, so should > either > > > give an error or actually union (or duplicate) elements... > > > > > > Stephan > > > > > > > > > On Wed, Nov 25, 2015 at 10:42 AM, Gyula Fóra > wrote: > > > > > >> Yes, I am not sure if this the intentional behaviour. I think you are > > >> supposed to be able to do the things you described. > > >> > > >> stream.union(stream.map(..)) and things like this are fair operations. > > Also > > >> maybe stream.union(stream) should just give stream instead of an > error. > > >> > > >> Could someone comment on this who knows the reasoning behind the > current > > >> mechanics? > > >> > > >> Gyula > > >> > > >> Vasiliki Kalavri ezt írta (időpont: 2015. > > nov. > > >> 24., K, 16:46): > > >> > > >>> Hi squirrels, > > >>> > > >>> when porting the gelly streaming code from 0.9 to 0.10 today with > > Paris, > > >> we > > >>> hit an exception in union: "*A DataStream cannot be unioned with > > >> itself*". > > >>> > > >>> The code raising this exception looks like this: > > >>> stream.union(stream.map(...)). > > >>> > > >>> Taking a look into the union code, we see that it's now not allowed > to > > >>> union a stream, not only with itself, but with any product of itself. > > >>> > > >>> First, we are wondering, why is that? Does it make building the > stream > > >>> graph easier in some way? > > >>> Second, we might want to give a better error message there, e.g. "*A > > >>> DataStream cannot be unioned with itself or a product of itself*", > and > > >>> finally, we should update the docs, which currently state that union > a > > >>> stream with itself is allowed and that "*If you union a data stream > > with > > >>> itself you will still only get each element once.*" > > >>> > > >>> Cheers, > > >>> -Vasia. > > >>> > > >> > > > > >
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
@Till I think the avro test data file is okay, the "no binaries" policy refers to binary executables, as far as I know. On Wed, Nov 25, 2015 at 2:54 PM, Till Rohrmannwrote: > Checked checksums for src release and Hadoop 2.7 Scala 2.10 release > > Checked binaries in source release > - contains ./flink-staging/flink-avro/src/test/resources/testdata.avro > > License > - no new files added which are relevant for licensing > > Build Flink and run tests from source release for Hadoop 2.5.1 > > Checked empty that log files don't contain exceptions and out files are > empty > > Run all examples with Hadoop 2.7 Scala 2.10 binaries via FliRTT tool on 4 > node standalone cluster and YARN cluster > > Tested planVisualizer > > Tested flink command line client > - tested info command > - tested -p option > > Tested cluster HA in standalone mode => working > > Tested cluster HA on YARN (2.7.1) => working > > Except for the avro testdata file which is contained in the source release, > I didn't find anything. > > +1 for releasing and removing the testdata file for the next release. > > On Wed, Nov 25, 2015 at 2:33 PM, Robert Metzger > wrote: > > > +1 > > > > - Build a maven project with the staging repository > > - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster > with > > YARN and HDFS HA > > - ran some kafka (0.8.2.0) read / write experiments > > - job cancellation with yarn is working ;) > > > > I found the following issue while testing: > > https://issues.apache.org/jira/browse/FLINK-3078 but it was already in > > 0.10.0 and its not super critical bc the JobManager container will be > > killed by YARN after a few minutes. > > > > > > I'll extend the vote until tomorrow Thursday, November 26. > > > > > > On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewen wrote: > > > > > @Gyula: I think it affects users, so should definitely be fixed very > soon > > > (either 0.10.1 or 0.10.2) > > > > > > Still checking whether Robert's current version fix solves it now, or > > > not... > > > > > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > > > vyacheslav.zholu...@gmail.com> wrote: > > > > > > > I can confirm that the build works fine when increasing a max number > of > > > > opened files. Sorry for confusion. > > > > > > > > > > > > > > > > -- > > > > View this message in context: > > > > > > > > > > http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html > > > > Sent from the Apache Flink Mailing List archive. mailing list archive > > at > > > > Nabble.com. > > > > > > > > > >
Re: Union a data stream with a product of itself
Hi, the operation “stream.union(stream.map(id))” is equivalent to “stream.union(stream)” isn’t it? So it might also duplicate the data. - Christoph > On 25 Nov 2015, at 11:24, Stephan Ewenwrote: > > "stream.union(stream.map(..))" should definitely be possible. Not sure why > this is not permitted. > > "stream.union(stream)" would contain each element twice, so should either > give an error or actually union (or duplicate) elements... > > Stephan > > > On Wed, Nov 25, 2015 at 10:42 AM, Gyula Fóra wrote: > >> Yes, I am not sure if this the intentional behaviour. I think you are >> supposed to be able to do the things you described. >> >> stream.union(stream.map(..)) and things like this are fair operations. Also >> maybe stream.union(stream) should just give stream instead of an error. >> >> Could someone comment on this who knows the reasoning behind the current >> mechanics? >> >> Gyula >> >> Vasiliki Kalavri ezt írta (időpont: 2015. nov. >> 24., K, 16:46): >> >>> Hi squirrels, >>> >>> when porting the gelly streaming code from 0.9 to 0.10 today with Paris, >> we >>> hit an exception in union: "*A DataStream cannot be unioned with >> itself*". >>> >>> The code raising this exception looks like this: >>> stream.union(stream.map(...)). >>> >>> Taking a look into the union code, we see that it's now not allowed to >>> union a stream, not only with itself, but with any product of itself. >>> >>> First, we are wondering, why is that? Does it make building the stream >>> graph easier in some way? >>> Second, we might want to give a better error message there, e.g. "*A >>> DataStream cannot be unioned with itself or a product of itself*", and >>> finally, we should update the docs, which currently state that union a >>> stream with itself is allowed and that "*If you union a data stream with >>> itself you will still only get each element once.*" >>> >>> Cheers, >>> -Vasia. >>> >>
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
+1 - License and Notice are good - ran all tests (including manual tests) work for hadoop 2.3.0 - Scala 2.10 - ran all tests for hadoop 2.7.0 - Scala 2.11 - ran all examples, several on larger external data - checked web frontend - checked quickstart archetypes On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewenwrote: > @Gyula: I think it affects users, so should definitely be fixed very soon > (either 0.10.1 or 0.10.2) > > Still checking whether Robert's current version fix solves it now, or > not... > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > vyacheslav.zholu...@gmail.com> wrote: > >> I can confirm that the build works fine when increasing a max number of >> opened files. Sorry for confusion. >> >> >> >> -- >> View this message in context: >> http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html >> Sent from the Apache Flink Mailing List archive. mailing list archive at >> Nabble.com. >> > >
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
Checked checksums for src release and Hadoop 2.7 Scala 2.10 release Checked binaries in source release - contains ./flink-staging/flink-avro/src/test/resources/testdata.avro License - no new files added which are relevant for licensing Build Flink and run tests from source release for Hadoop 2.5.1 Checked empty that log files don't contain exceptions and out files are empty Run all examples with Hadoop 2.7 Scala 2.10 binaries via FliRTT tool on 4 node standalone cluster and YARN cluster Tested planVisualizer Tested flink command line client - tested info command - tested -p option Tested cluster HA in standalone mode => working Tested cluster HA on YARN (2.7.1) => working Except for the avro testdata file which is contained in the source release, I didn't find anything. +1 for releasing and removing the testdata file for the next release. On Wed, Nov 25, 2015 at 2:33 PM, Robert Metzgerwrote: > +1 > > - Build a maven project with the staging repository > - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster with > YARN and HDFS HA > - ran some kafka (0.8.2.0) read / write experiments > - job cancellation with yarn is working ;) > > I found the following issue while testing: > https://issues.apache.org/jira/browse/FLINK-3078 but it was already in > 0.10.0 and its not super critical bc the JobManager container will be > killed by YARN after a few minutes. > > > I'll extend the vote until tomorrow Thursday, November 26. > > > On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewen wrote: > > > @Gyula: I think it affects users, so should definitely be fixed very soon > > (either 0.10.1 or 0.10.2) > > > > Still checking whether Robert's current version fix solves it now, or > > not... > > > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > > vyacheslav.zholu...@gmail.com> wrote: > > > > > I can confirm that the build works fine when increasing a max number of > > > opened files. Sorry for confusion. > > > > > > > > > > > > -- > > > View this message in context: > > > > > > http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html > > > Sent from the Apache Flink Mailing List archive. mailing list archive > at > > > Nabble.com. > > > > > >
Re: [VOTE] Release Apache Flink 0.10.1 (release-0.10.0-rc1)
Alright, then I withdraw my remark concerning testdata.avro. On Wed, Nov 25, 2015 at 2:56 PM, Stephan Ewenwrote: > @Till I think the avro test data file is okay, the "no binaries" policy > refers to binary executables, as far as I know. > > On Wed, Nov 25, 2015 at 2:54 PM, Till Rohrmann > wrote: > > > Checked checksums for src release and Hadoop 2.7 Scala 2.10 release > > > > Checked binaries in source release > > - contains ./flink-staging/flink-avro/src/test/resources/testdata.avro > > > > License > > - no new files added which are relevant for licensing > > > > Build Flink and run tests from source release for Hadoop 2.5.1 > > > > Checked empty that log files don't contain exceptions and out files are > > empty > > > > Run all examples with Hadoop 2.7 Scala 2.10 binaries via FliRTT tool on 4 > > node standalone cluster and YARN cluster > > > > Tested planVisualizer > > > > Tested flink command line client > > - tested info command > > - tested -p option > > > > Tested cluster HA in standalone mode => working > > > > Tested cluster HA on YARN (2.7.1) => working > > > > Except for the avro testdata file which is contained in the source > release, > > I didn't find anything. > > > > +1 for releasing and removing the testdata file for the next release. > > > > On Wed, Nov 25, 2015 at 2:33 PM, Robert Metzger > > wrote: > > > > > +1 > > > > > > - Build a maven project with the staging repository > > > - started Flink on YARN on a CDH 5.4.5 / Hadoop 2.6.0-cdh5.4.5 cluster > > with > > > YARN and HDFS HA > > > - ran some kafka (0.8.2.0) read / write experiments > > > - job cancellation with yarn is working ;) > > > > > > I found the following issue while testing: > > > https://issues.apache.org/jira/browse/FLINK-3078 but it was already in > > > 0.10.0 and its not super critical bc the JobManager container will be > > > killed by YARN after a few minutes. > > > > > > > > > I'll extend the vote until tomorrow Thursday, November 26. > > > > > > > > > On Tue, Nov 24, 2015 at 1:54 PM, Stephan Ewen > wrote: > > > > > > > @Gyula: I think it affects users, so should definitely be fixed very > > soon > > > > (either 0.10.1 or 0.10.2) > > > > > > > > Still checking whether Robert's current version fix solves it now, or > > > > not... > > > > > > > > On Tue, Nov 24, 2015 at 1:46 PM, Vyacheslav Zholudev < > > > > vyacheslav.zholu...@gmail.com> wrote: > > > > > > > > > I can confirm that the build works fine when increasing a max > number > > of > > > > > opened files. Sorry for confusion. > > > > > > > > > > > > > > > > > > > > -- > > > > > View this message in context: > > > > > > > > > > > > > > > http://apache-flink-mailing-list-archive.1008284.n3.nabble.com/VOTE-Release-Apache-Flink-0-10-1-release-0-10-0-rc1-tp9296p9327.html > > > > > Sent from the Apache Flink Mailing List archive. mailing list > archive > > > at > > > > > Nabble.com. > > > > > > > > > > > > > > >
RE: The null in Flink
Thanks, Timo. We may put the NULL related function support to SQL API, but for Scalar expression and Boolean expression, it already been supported in Table API, without NULL value handling support, query with Scalar expression and Boolean expression would fail while encounter NULL value. Thanks Chengxiang -Original Message- From: Timo Walther [mailto:twal...@apache.org] Sent: Wednesday, November 25, 2015 7:33 PM To: dev@flink.apache.org Subject: Re: The null in Flink Hi Chengxiang, I totally agree that the Table API should fully support NULL values. The Table API is a logical API and therefore we should be as close to ANSI SQL as possible. Rows need to be nullable in the near future. 2. i, ii, iii and iv sound reasonable. But v, vi and vii sound to much like SQL magic. I think all other SQL magic (DBMS specific corner cases) should be handled by the SQL API on top of the Table API. Regards, Timo On 25.11.2015 11:31, Li, Chengxiang wrote: > Hi > In this mail list, there are some discussions about null value handling in > Flink, and I saw several related JIRAs as well(like FLINK-2203, FLINK-2210), > but unfortunately, got reverted due to immature design, and no further action > since then. I would like to pick this topic up here, as it's quite an > important part of data analysis and many features depend on it. Hopefully, > through a plenary discussion, we can generate an acceptable solution and move > forward. Stephan has explained very clearly about how and why Flink handle > "Null values in the Programming Language APIs", so I mainly talk about the > second part of "Null values in the high-level (logical) APIs ". > > 1. Why should Flink support Null values handling in Table API? > i. Data source may miss column value in many cases, if no Null values > handling in Table API, user need to write an extra ETL to handle missing > values manually. > ii. Some Table API operators generate Null values on their own, like > Outer Join/Cube/Rollup/Grouping Set, and so on. Null values handling in Table > API is the prerequisite of these features. > > 2. The semantic of Null value handling in Table API. > Fortunately, there are already mature DBMS standards we can follow for Null > value handling, I list several semantic of Null value handling here. To be > noted that, this may not cover all the cases, and the semantics may vary in > different DBMSs, so it should totally open to discuss. > I, NULL compare. In ascending order, NULL is smaller than any other > value, and NULL == NULL return false. > ii. NULL exists in GroupBy Key, all NULL values are grouped as a single > group. > iii. NULL exists in Aggregate columns, ignore NULL in aggregation > function. > iv. NULL exists in both side Join key, refer to #i, NULL == > NULL return false, no output for NULL Join key. > v. NULL in Scalar expression, expression within NULL(eg. 1 > + NULL) return NULL. > vi. NULL in Boolean expression, add an extra result: > UNKNOWN, more semantic for Boolean expression in reference #1. > vii. More related function support, like COALESCE, NVL, > NANVL, and so on. > > 3. NULL value storage in Table API. >Just set null to Row field value. To mark NULL value in serialized binary > record data, normally it use extra flag for each field to mark whether its > value is NULL, which would change the data layout of Row object. So any logic > that access serialized Row data directly should updated to sync with new data > layout, for example, many methods in RowComparator. > > Reference: > 1. Nulls: Nothing to worry about: > http://www.oracle.com/technetwork/issue-archive/2005/05-jul/o45sql-097727.html. > 2. Null related functions: > https://oracle-base.com/articles/misc/null-related-functions > > -Original Message- > From: ewenstep...@gmail.com [mailto:ewenstep...@gmail.com] On Behalf > Of Stephan Ewen > Sent: Thursday, June 18, 2015 8:43 AM > To: dev@flink.apache.org > Subject: Re: The null in Flink > > Hi! > > I think we actually have two discussions here, both of them important: > > -- > 1) Null values in the Programming Language APIs > -- > > Fields in composite types may simply be null pointers. > > In object types: >- primitives members are naturally non-nullable >- all other members are nullable > > => If you want to avoid the overhead of nullability, go with primitive types. > > In Tuples, and derives types (Scala case classes): >- Fields are non-nullable. > > => The reason here is that we initially decided to keep tuples as a very fast > data type. Because tuples cannot hold primitives in Java/Scala, we would not > have a way to make fast non-nullable fields. The performance of nullable > fields affects the key-operations, especially on