[jira] [Created] (FLINK-2794) Exactly-once support
Maximilian Michels created FLINK-2794: - Summary: Exactly-once support Key: FLINK-2794 URL: https://issues.apache.org/jira/browse/FLINK-2794 Project: Flink Issue Type: Bug Components: Documentation, Streaming Reporter: Maximilian Michels -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: Pulling Streaming out of staging and project restructure
Great to see streaming graduating. :) I like the outline, both getting rid of staging, having the examples together and generally flattening the structure are very reasonable to me. You have listed flink-streaming-examples under flink-streaming-connectors and left out some less prominent maven modules, but I assume the first is accidental while the second is intentional to make the list a bit briefer. Best, Marton On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewenwrote: > Hi all! > > We are making good headway with reworking the last parts of the Window API. > After that, the streaming API should be good to be pulled out of staging. > > Since we are reorganizing the projects as part of that, I would shift a bit > more to bring things a bit more up to date. > > In this restructure, I would like to get rid of the "flink-staging" > project. Anyone who only uses the maven artifacts sees no difference > whether a project is in "staging" or not, so it does not help much to have > that directory structure. > On the other hand, projects have a tendency to linger in staging forever > (like avro, spargel, hbase, jdbc, ...) > > The new structure could be > > flink-core > flink-java > flink-scala > flink-streaming-core > flink-streaming-scala > > flink-runtime > flink-runtime-web > flink-optimizer > flink-clients > > flink-shaded > -> flink-shaded-hadoop > -> flink-shaded-hadoop2 > -> flink-shaded-include-yarn-tests > -> flink-shaded-curator > > flink-examples > -> (have all examples, Scala and Java, Batch and Streaming) > > flink-batch-connectors > -> flink-avro > -> flink-jdbc > -> flink-hadoop-compatibility > -> flink-hbase > -> flink-hcatalog > > flink-streaming-connectors > -> flink-connector-twitter > -> flink-streaming-examples > -> flink-connector-flume > -> flink-connector-kafka > -> flink-connector-elasticsearch > -> flink-connector-rabbitmq > -> flink-connector-filesystem > > flink-libraries > -> flink-gelly > -> flink-gelly-scala > -> flink-ml > -> flink-table > -> flink-language-binding > -> flink-python > > > flink-scala-shell > > flink-test-utils > flink-tests > flink-fs-tests > > flink-contrib > -> flink-storm-compatibility > -> flink-storm-compatibility-examples > -> flink-streaming-utils > -> flink-tweet-inputformat > -> flink-operator-stats > -> flink-tez > > flink-quickstart > -> flink-quickstart-java > -> flink-quickstart-scala > -> flink-tez-quickstart > > flink-yarn > flink-yarn-tests > > flink-dist > > flink-benchmark > > > Let me know if that makes sense! > > Greetings, > Stephan >
Re: Pulling Streaming out of staging and project restructure
+1 for Robert's comments. On Thu, Oct 1, 2015 at 3:16 PM, Robert Metzgerwrote: > Big +1 for graduating streaming out of staging. It is widely used, also in > production and we are spending a lot of effort into hardening it. > I also agree with the proposed new maven module structure. > > We have to carefully test the reworked structure for the scripts which are > generating the hadoop1 and the scala 2.11 poms (they are transformed using > a bunch of bash scripts). I can do that once the PR is open. > > @Chesnay: I would be fine with including the language binding into python > > Where would new projects reside in, that previously would have been put > into flink-staging? > > flink-contrib > > > @Kostas: I understand the idea behind your suggested renaming, but thats > just a name. I don't think its going to influence how people are seeing > Flink: It doesn't feel like second class when adding "flink-streaming-core" > to the dependencies to me. > Also, the "flink-datastream-scala" module would depend on > "flink-dataset-scala", which is kind of weird. > > > I'm wondering whether we should remove the "flink-test-utils" module. I > don't think its really necessary, because we can put the test jars into the > flink-tests project and include them using the "test-jar" dependency. > > > On Thu, Oct 1, 2015 at 2:27 PM, Kostas Tzoumas > wrote: > > > +1 > > > > I wanted to suggest that we rename modules to fully accept streaming as > > first class, qualifying also "batch" as "batch" (e.g., flink-java --> > > flink-dataset-java, flink-streaming --> flink-datastream, etc). > > > > This would break maven dependencies (temporary hell :-) so it's not a > > decision to take lightly. I'm not strongly advocating for it. > > > > > > On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Schepler > > wrote: > > > > > I like it in general. But while we're at it, what is the purpose of the > > > flink-tests project, or rather which tests belong there instead of the > > > individual projects? > > > > > > Where would new projects reside in, that previously would have been put > > > into flink-staging? > > > > > > Lastly, I'd like to merge flink-language-binding into flink-python. I > can > > > go more into detail but the gist of it is that the abstraction just > > doesn't > > > work. > > > > > > > > > On 01.10.2015 12:40, Márton Balassi wrote: > > > > > >> Great to see streaming graduating. :) > > >> > > >> I like the outline, both getting rid of staging, having the examples > > >> together and generally flattening the structure are very reasonable to > > me. > > >> > > >> You have listed flink-streaming-examples under > > flink-streaming-connectors > > >> and left out some less prominent maven modules, but I assume the first > > is > > >> accidental while the second is intentional to make the list a bit > > briefer. > > >> > > >> Best, > > >> > > >> Marton > > >> > > >> > > >> On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen > wrote: > > >> > > >> Hi all! > > >>> > > >>> We are making good headway with reworking the last parts of the > Window > > >>> API. > > >>> After that, the streaming API should be good to be pulled out of > > staging. > > >>> > > >>> Since we are reorganizing the projects as part of that, I would > shift a > > >>> bit > > >>> more to bring things a bit more up to date. > > >>> > > >>> In this restructure, I would like to get rid of the "flink-staging" > > >>> project. Anyone who only uses the maven artifacts sees no difference > > >>> whether a project is in "staging" or not, so it does not help much to > > >>> have > > >>> that directory structure. > > >>> On the other hand, projects have a tendency to linger in staging > > forever > > >>> (like avro, spargel, hbase, jdbc, ...) > > >>> > > >>> The new structure could be > > >>> > > >>> flink-core > > >>> flink-java > > >>> flink-scala > > >>> flink-streaming-core > > >>> flink-streaming-scala > > >>> > > >>> flink-runtime > > >>> flink-runtime-web > > >>> flink-optimizer > > >>> flink-clients > > >>> > > >>> flink-shaded > > >>>-> flink-shaded-hadoop > > >>>-> flink-shaded-hadoop2 > > >>>-> flink-shaded-include-yarn-tests > > >>>-> flink-shaded-curator > > >>> > > >>> flink-examples > > >>>-> (have all examples, Scala and Java, Batch and Streaming) > > >>> > > >>> flink-batch-connectors > > >>>-> flink-avro > > >>>-> flink-jdbc > > >>>-> flink-hadoop-compatibility > > >>>-> flink-hbase > > >>>-> flink-hcatalog > > >>> > > >>> flink-streaming-connectors > > >>>-> flink-connector-twitter > > >>>-> flink-streaming-examples > > >>>-> flink-connector-flume > > >>>-> flink-connector-kafka > > >>>-> flink-connector-elasticsearch > > >>>-> flink-connector-rabbitmq > > >>>-> flink-connector-filesystem > > >>> > > >>> flink-libraries > > >>>-> flink-gelly > > >>>-> flink-gelly-scala > > >>>-> flink-ml > > >>>->
Re: Pulling Streaming out of staging and project restructure
If we remove flink-staging because projects tend to get stuck there, what mechanism prevents the same happening with flink-contrib? On 01.10.2015 15:19, Stephan Ewen wrote: +1 for Robert's comments. On Thu, Oct 1, 2015 at 3:16 PM, Robert Metzgerwrote: Big +1 for graduating streaming out of staging. It is widely used, also in production and we are spending a lot of effort into hardening it. I also agree with the proposed new maven module structure. We have to carefully test the reworked structure for the scripts which are generating the hadoop1 and the scala 2.11 poms (they are transformed using a bunch of bash scripts). I can do that once the PR is open. @Chesnay: I would be fine with including the language binding into python Where would new projects reside in, that previously would have been put into flink-staging? flink-contrib @Kostas: I understand the idea behind your suggested renaming, but thats just a name. I don't think its going to influence how people are seeing Flink: It doesn't feel like second class when adding "flink-streaming-core" to the dependencies to me. Also, the "flink-datastream-scala" module would depend on "flink-dataset-scala", which is kind of weird. I'm wondering whether we should remove the "flink-test-utils" module. I don't think its really necessary, because we can put the test jars into the flink-tests project and include them using the "test-jar" dependency. On Thu, Oct 1, 2015 at 2:27 PM, Kostas Tzoumas wrote: +1 I wanted to suggest that we rename modules to fully accept streaming as first class, qualifying also "batch" as "batch" (e.g., flink-java --> flink-dataset-java, flink-streaming --> flink-datastream, etc). This would break maven dependencies (temporary hell :-) so it's not a decision to take lightly. I'm not strongly advocating for it. On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Schepler wrote: I like it in general. But while we're at it, what is the purpose of the flink-tests project, or rather which tests belong there instead of the individual projects? Where would new projects reside in, that previously would have been put into flink-staging? Lastly, I'd like to merge flink-language-binding into flink-python. I can go more into detail but the gist of it is that the abstraction just doesn't work. On 01.10.2015 12:40, Márton Balassi wrote: Great to see streaming graduating. :) I like the outline, both getting rid of staging, having the examples together and generally flattening the structure are very reasonable to me. You have listed flink-streaming-examples under flink-streaming-connectors and left out some less prominent maven modules, but I assume the first is accidental while the second is intentional to make the list a bit briefer. Best, Marton On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen wrote: Hi all! We are making good headway with reworking the last parts of the Window API. After that, the streaming API should be good to be pulled out of staging. Since we are reorganizing the projects as part of that, I would shift a bit more to bring things a bit more up to date. In this restructure, I would like to get rid of the "flink-staging" project. Anyone who only uses the maven artifacts sees no difference whether a project is in "staging" or not, so it does not help much to have that directory structure. On the other hand, projects have a tendency to linger in staging forever (like avro, spargel, hbase, jdbc, ...) The new structure could be flink-core flink-java flink-scala flink-streaming-core flink-streaming-scala flink-runtime flink-runtime-web flink-optimizer flink-clients flink-shaded -> flink-shaded-hadoop -> flink-shaded-hadoop2 -> flink-shaded-include-yarn-tests -> flink-shaded-curator flink-examples -> (have all examples, Scala and Java, Batch and Streaming) flink-batch-connectors -> flink-avro -> flink-jdbc -> flink-hadoop-compatibility -> flink-hbase -> flink-hcatalog flink-streaming-connectors -> flink-connector-twitter -> flink-streaming-examples -> flink-connector-flume -> flink-connector-kafka -> flink-connector-elasticsearch -> flink-connector-rabbitmq -> flink-connector-filesystem flink-libraries -> flink-gelly -> flink-gelly-scala -> flink-ml -> flink-table -> flink-language-binding -> flink-python flink-scala-shell flink-test-utils flink-tests flink-fs-tests flink-contrib -> flink-storm-compatibility -> flink-storm-compatibility-examples -> flink-streaming-utils -> flink-tweet-inputformat -> flink-operator-stats -> flink-tez flink-quickstart -> flink-quickstart-java -> flink-quickstart-scala -> flink-tez-quickstart flink-yarn flink-yarn-tests flink-dist flink-benchmark Let me know if that makes sense! Greetings, Stephan
[jira] [Created] (FLINK-2795) Print JobExecutionResult for interactively invoked jobs
Maximilian Michels created FLINK-2795: - Summary: Print JobExecutionResult for interactively invoked jobs Key: FLINK-2795 URL: https://issues.apache.org/jira/browse/FLINK-2795 Project: Flink Issue Type: Improvement Components: Command-line client Affects Versions: 0.9, 0.10 Reporter: Maximilian Michels Fix For: 0.10 {{JobExecutionResult}}s are currently not available if the Flink job is submitted interactively via the command-line client. The reason for that is that the execution goes through the {{ContextEnvironment}} which itself holds a reference to the {{Client}} which it uses to submit the job. We need to pass the JobExecutionResult back to the Client like we do it for the JobID. -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Pulling Streaming out of staging and project restructure
Hi all! We are making good headway with reworking the last parts of the Window API. After that, the streaming API should be good to be pulled out of staging. Since we are reorganizing the projects as part of that, I would shift a bit more to bring things a bit more up to date. In this restructure, I would like to get rid of the "flink-staging" project. Anyone who only uses the maven artifacts sees no difference whether a project is in "staging" or not, so it does not help much to have that directory structure. On the other hand, projects have a tendency to linger in staging forever (like avro, spargel, hbase, jdbc, ...) The new structure could be flink-core flink-java flink-scala flink-streaming-core flink-streaming-scala flink-runtime flink-runtime-web flink-optimizer flink-clients flink-shaded -> flink-shaded-hadoop -> flink-shaded-hadoop2 -> flink-shaded-include-yarn-tests -> flink-shaded-curator flink-examples -> (have all examples, Scala and Java, Batch and Streaming) flink-batch-connectors -> flink-avro -> flink-jdbc -> flink-hadoop-compatibility -> flink-hbase -> flink-hcatalog flink-streaming-connectors -> flink-connector-twitter -> flink-streaming-examples -> flink-connector-flume -> flink-connector-kafka -> flink-connector-elasticsearch -> flink-connector-rabbitmq -> flink-connector-filesystem flink-libraries -> flink-gelly -> flink-gelly-scala -> flink-ml -> flink-table -> flink-language-binding -> flink-python flink-scala-shell flink-test-utils flink-tests flink-fs-tests flink-contrib -> flink-storm-compatibility -> flink-storm-compatibility-examples -> flink-streaming-utils -> flink-tweet-inputformat -> flink-operator-stats -> flink-tez flink-quickstart -> flink-quickstart-java -> flink-quickstart-scala -> flink-tez-quickstart flink-yarn flink-yarn-tests flink-dist flink-benchmark Let me know if that makes sense! Greetings, Stephan
Re: Pulling Streaming out of staging and project restructure
I like it in general. But while we're at it, what is the purpose of the flink-tests project, or rather which tests belong there instead of the individual projects? Where would new projects reside in, that previously would have been put into flink-staging? Lastly, I'd like to merge flink-language-binding into flink-python. I can go more into detail but the gist of it is that the abstraction just doesn't work. On 01.10.2015 12:40, Márton Balassi wrote: Great to see streaming graduating. :) I like the outline, both getting rid of staging, having the examples together and generally flattening the structure are very reasonable to me. You have listed flink-streaming-examples under flink-streaming-connectors and left out some less prominent maven modules, but I assume the first is accidental while the second is intentional to make the list a bit briefer. Best, Marton On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewenwrote: Hi all! We are making good headway with reworking the last parts of the Window API. After that, the streaming API should be good to be pulled out of staging. Since we are reorganizing the projects as part of that, I would shift a bit more to bring things a bit more up to date. In this restructure, I would like to get rid of the "flink-staging" project. Anyone who only uses the maven artifacts sees no difference whether a project is in "staging" or not, so it does not help much to have that directory structure. On the other hand, projects have a tendency to linger in staging forever (like avro, spargel, hbase, jdbc, ...) The new structure could be flink-core flink-java flink-scala flink-streaming-core flink-streaming-scala flink-runtime flink-runtime-web flink-optimizer flink-clients flink-shaded -> flink-shaded-hadoop -> flink-shaded-hadoop2 -> flink-shaded-include-yarn-tests -> flink-shaded-curator flink-examples -> (have all examples, Scala and Java, Batch and Streaming) flink-batch-connectors -> flink-avro -> flink-jdbc -> flink-hadoop-compatibility -> flink-hbase -> flink-hcatalog flink-streaming-connectors -> flink-connector-twitter -> flink-streaming-examples -> flink-connector-flume -> flink-connector-kafka -> flink-connector-elasticsearch -> flink-connector-rabbitmq -> flink-connector-filesystem flink-libraries -> flink-gelly -> flink-gelly-scala -> flink-ml -> flink-table -> flink-language-binding -> flink-python flink-scala-shell flink-test-utils flink-tests flink-fs-tests flink-contrib -> flink-storm-compatibility -> flink-storm-compatibility-examples -> flink-streaming-utils -> flink-tweet-inputformat -> flink-operator-stats -> flink-tez flink-quickstart -> flink-quickstart-java -> flink-quickstart-scala -> flink-tez-quickstart flink-yarn flink-yarn-tests flink-dist flink-benchmark Let me know if that makes sense! Greetings, Stephan
Re: Pulling Streaming out of staging and project restructure
+1 I wanted to suggest that we rename modules to fully accept streaming as first class, qualifying also "batch" as "batch" (e.g., flink-java --> flink-dataset-java, flink-streaming --> flink-datastream, etc). This would break maven dependencies (temporary hell :-) so it's not a decision to take lightly. I'm not strongly advocating for it. On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Scheplerwrote: > I like it in general. But while we're at it, what is the purpose of the > flink-tests project, or rather which tests belong there instead of the > individual projects? > > Where would new projects reside in, that previously would have been put > into flink-staging? > > Lastly, I'd like to merge flink-language-binding into flink-python. I can > go more into detail but the gist of it is that the abstraction just doesn't > work. > > > On 01.10.2015 12:40, Márton Balassi wrote: > >> Great to see streaming graduating. :) >> >> I like the outline, both getting rid of staging, having the examples >> together and generally flattening the structure are very reasonable to me. >> >> You have listed flink-streaming-examples under flink-streaming-connectors >> and left out some less prominent maven modules, but I assume the first is >> accidental while the second is intentional to make the list a bit briefer. >> >> Best, >> >> Marton >> >> >> On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen wrote: >> >> Hi all! >>> >>> We are making good headway with reworking the last parts of the Window >>> API. >>> After that, the streaming API should be good to be pulled out of staging. >>> >>> Since we are reorganizing the projects as part of that, I would shift a >>> bit >>> more to bring things a bit more up to date. >>> >>> In this restructure, I would like to get rid of the "flink-staging" >>> project. Anyone who only uses the maven artifacts sees no difference >>> whether a project is in "staging" or not, so it does not help much to >>> have >>> that directory structure. >>> On the other hand, projects have a tendency to linger in staging forever >>> (like avro, spargel, hbase, jdbc, ...) >>> >>> The new structure could be >>> >>> flink-core >>> flink-java >>> flink-scala >>> flink-streaming-core >>> flink-streaming-scala >>> >>> flink-runtime >>> flink-runtime-web >>> flink-optimizer >>> flink-clients >>> >>> flink-shaded >>>-> flink-shaded-hadoop >>>-> flink-shaded-hadoop2 >>>-> flink-shaded-include-yarn-tests >>>-> flink-shaded-curator >>> >>> flink-examples >>>-> (have all examples, Scala and Java, Batch and Streaming) >>> >>> flink-batch-connectors >>>-> flink-avro >>>-> flink-jdbc >>>-> flink-hadoop-compatibility >>>-> flink-hbase >>>-> flink-hcatalog >>> >>> flink-streaming-connectors >>>-> flink-connector-twitter >>>-> flink-streaming-examples >>>-> flink-connector-flume >>>-> flink-connector-kafka >>>-> flink-connector-elasticsearch >>>-> flink-connector-rabbitmq >>>-> flink-connector-filesystem >>> >>> flink-libraries >>>-> flink-gelly >>>-> flink-gelly-scala >>>-> flink-ml >>>-> flink-table >>>-> flink-language-binding >>>-> flink-python >>> >>> >>> flink-scala-shell >>> >>> flink-test-utils >>> flink-tests >>> flink-fs-tests >>> >>> flink-contrib >>>-> flink-storm-compatibility >>>-> flink-storm-compatibility-examples >>>-> flink-streaming-utils >>>-> flink-tweet-inputformat >>>-> flink-operator-stats >>>-> flink-tez >>> >>> flink-quickstart >>>-> flink-quickstart-java >>>-> flink-quickstart-scala >>>-> flink-tez-quickstart >>> >>> flink-yarn >>> flink-yarn-tests >>> >>> flink-dist >>> >>> flink-benchmark >>> >>> >>> Let me know if that makes sense! >>> >>> Greetings, >>> Stephan >>> >>> >
Re: Pulling Streaming out of staging and project restructure
Big +1 for graduating streaming out of staging. It is widely used, also in production and we are spending a lot of effort into hardening it. I also agree with the proposed new maven module structure. We have to carefully test the reworked structure for the scripts which are generating the hadoop1 and the scala 2.11 poms (they are transformed using a bunch of bash scripts). I can do that once the PR is open. @Chesnay: I would be fine with including the language binding into python > Where would new projects reside in, that previously would have been put into flink-staging? flink-contrib @Kostas: I understand the idea behind your suggested renaming, but thats just a name. I don't think its going to influence how people are seeing Flink: It doesn't feel like second class when adding "flink-streaming-core" to the dependencies to me. Also, the "flink-datastream-scala" module would depend on "flink-dataset-scala", which is kind of weird. I'm wondering whether we should remove the "flink-test-utils" module. I don't think its really necessary, because we can put the test jars into the flink-tests project and include them using the "test-jar" dependency. On Thu, Oct 1, 2015 at 2:27 PM, Kostas Tzoumaswrote: > +1 > > I wanted to suggest that we rename modules to fully accept streaming as > first class, qualifying also "batch" as "batch" (e.g., flink-java --> > flink-dataset-java, flink-streaming --> flink-datastream, etc). > > This would break maven dependencies (temporary hell :-) so it's not a > decision to take lightly. I'm not strongly advocating for it. > > > On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Schepler > wrote: > > > I like it in general. But while we're at it, what is the purpose of the > > flink-tests project, or rather which tests belong there instead of the > > individual projects? > > > > Where would new projects reside in, that previously would have been put > > into flink-staging? > > > > Lastly, I'd like to merge flink-language-binding into flink-python. I can > > go more into detail but the gist of it is that the abstraction just > doesn't > > work. > > > > > > On 01.10.2015 12:40, Márton Balassi wrote: > > > >> Great to see streaming graduating. :) > >> > >> I like the outline, both getting rid of staging, having the examples > >> together and generally flattening the structure are very reasonable to > me. > >> > >> You have listed flink-streaming-examples under > flink-streaming-connectors > >> and left out some less prominent maven modules, but I assume the first > is > >> accidental while the second is intentional to make the list a bit > briefer. > >> > >> Best, > >> > >> Marton > >> > >> > >> On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen wrote: > >> > >> Hi all! > >>> > >>> We are making good headway with reworking the last parts of the Window > >>> API. > >>> After that, the streaming API should be good to be pulled out of > staging. > >>> > >>> Since we are reorganizing the projects as part of that, I would shift a > >>> bit > >>> more to bring things a bit more up to date. > >>> > >>> In this restructure, I would like to get rid of the "flink-staging" > >>> project. Anyone who only uses the maven artifacts sees no difference > >>> whether a project is in "staging" or not, so it does not help much to > >>> have > >>> that directory structure. > >>> On the other hand, projects have a tendency to linger in staging > forever > >>> (like avro, spargel, hbase, jdbc, ...) > >>> > >>> The new structure could be > >>> > >>> flink-core > >>> flink-java > >>> flink-scala > >>> flink-streaming-core > >>> flink-streaming-scala > >>> > >>> flink-runtime > >>> flink-runtime-web > >>> flink-optimizer > >>> flink-clients > >>> > >>> flink-shaded > >>>-> flink-shaded-hadoop > >>>-> flink-shaded-hadoop2 > >>>-> flink-shaded-include-yarn-tests > >>>-> flink-shaded-curator > >>> > >>> flink-examples > >>>-> (have all examples, Scala and Java, Batch and Streaming) > >>> > >>> flink-batch-connectors > >>>-> flink-avro > >>>-> flink-jdbc > >>>-> flink-hadoop-compatibility > >>>-> flink-hbase > >>>-> flink-hcatalog > >>> > >>> flink-streaming-connectors > >>>-> flink-connector-twitter > >>>-> flink-streaming-examples > >>>-> flink-connector-flume > >>>-> flink-connector-kafka > >>>-> flink-connector-elasticsearch > >>>-> flink-connector-rabbitmq > >>>-> flink-connector-filesystem > >>> > >>> flink-libraries > >>>-> flink-gelly > >>>-> flink-gelly-scala > >>>-> flink-ml > >>>-> flink-table > >>>-> flink-language-binding > >>>-> flink-python > >>> > >>> > >>> flink-scala-shell > >>> > >>> flink-test-utils > >>> flink-tests > >>> flink-fs-tests > >>> > >>> flink-contrib > >>>-> flink-storm-compatibility > >>>-> flink-storm-compatibility-examples > >>>-> flink-streaming-utils > >>>-> flink-tweet-inputformat > >>>->
[jira] [Created] (FLINK-2798) Prepare new web dashboard for executing in on YARN
Robert Metzger created FLINK-2798: - Summary: Prepare new web dashboard for executing in on YARN Key: FLINK-2798 URL: https://issues.apache.org/jira/browse/FLINK-2798 Project: Flink Issue Type: Bug Components: Webfrontend, YARN Client Reporter: Robert Metzger -- This message was sent by Atlassian JIRA (v6.3.4#6332)
Re: Pulling Streaming out of staging and project restructure
@Chesnay: Nothing prevents projects from getting stuck there. But at least we don't have two staging repositories (dist, staging). Also I would not say that there has been no graduation out of staging. Yarn was also there once, streaming and gelly are currently leaving it. So I would actually say it worked. One more thing: Can we actually put the web submission client into a separate module until we have something new and better? It does really not fit into "flink-clients". It makes Flink users pull the jetty dependencies even if they never use them for example when running Flink from an IDE. On Thu, Oct 1, 2015 at 4:45 PM, Kostas Tzoumaswrote: > +1 to Robert and practicality :-) > > As I said before, I do not feel strongly about this, I was torn myself. > > On Thu, Oct 1, 2015 at 3:28 PM, Chesnay Schepler > wrote: > > > If we remove flink-staging because projects tend to get stuck there, what > > mechanism prevents the same happening with flink-contrib? > > > > > > On 01.10.2015 15:19, Stephan Ewen wrote: > > > >> +1 for Robert's comments. > >> > >> On Thu, Oct 1, 2015 at 3:16 PM, Robert Metzger > >> wrote: > >> > >> Big +1 for graduating streaming out of staging. It is widely used, also > in > >>> production and we are spending a lot of effort into hardening it. > >>> I also agree with the proposed new maven module structure. > >>> > >>> We have to carefully test the reworked structure for the scripts which > >>> are > >>> generating the hadoop1 and the scala 2.11 poms (they are transformed > >>> using > >>> a bunch of bash scripts). I can do that once the PR is open. > >>> > >>> @Chesnay: I would be fine with including the language binding into > python > >>> > Where would new projects reside in, that previously would have been > put > > >>> into flink-staging? > >>> > >>> flink-contrib > >>> > >>> > >>> @Kostas: I understand the idea behind your suggested renaming, but > thats > >>> just a name. I don't think its going to influence how people are seeing > >>> Flink: It doesn't feel like second class when adding > >>> "flink-streaming-core" > >>> to the dependencies to me. > >>> Also, the "flink-datastream-scala" module would depend on > >>> "flink-dataset-scala", which is kind of weird. > >>> > >>> > >>> I'm wondering whether we should remove the "flink-test-utils" module. I > >>> don't think its really necessary, because we can put the test jars into > >>> the > >>> flink-tests project and include them using the "test-jar" dependency. > >>> > >>> > >>> On Thu, Oct 1, 2015 at 2:27 PM, Kostas Tzoumas > >>> wrote: > >>> > >>> +1 > > I wanted to suggest that we rename modules to fully accept streaming > as > first class, qualifying also "batch" as "batch" (e.g., flink-java --> > flink-dataset-java, flink-streaming --> flink-datastream, etc). > > This would break maven dependencies (temporary hell :-) so it's not a > decision to take lightly. I'm not strongly advocating for it. > > > On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Schepler > > wrote: > > I like it in general. But while we're at it, what is the purpose of > the > > flink-tests project, or rather which tests belong there instead of > the > > individual projects? > > > > Where would new projects reside in, that previously would have been > put > > into flink-staging? > > > > Lastly, I'd like to merge flink-language-binding into flink-python. I > > > can > >>> > go more into detail but the gist of it is that the abstraction just > > > doesn't > > > work. > > > > > > On 01.10.2015 12:40, Márton Balassi wrote: > > > > Great to see streaming graduating. :) > >> > >> I like the outline, both getting rid of staging, having the examples > >> together and generally flattening the structure are very reasonable > to > >> > > me. > > > You have listed flink-streaming-examples under > >> > > flink-streaming-connectors > > > and left out some less prominent maven modules, but I assume the > first > >> > > is > > > accidental while the second is intentional to make the list a bit > >> > > briefer. > > > Best, > >> > >> Marton > >> > >> > >> On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen > >> > > wrote: > >>> > Hi all! > >> > >>> We are making good headway with reworking the last parts of the > >>> > >> Window > >>> > API. > >>> After that, the streaming API should be good to be pulled out of > >>> > >> staging. > > > Since we are reorganizing the projects as part of that, I would > >>> > >> shift a > >>> > bit > >>> more to bring things a bit more up to date. > >>> > >>> In this restructure, I would
Re: Pulling Streaming out of staging and project restructure
+1 to Robert and practicality :-) As I said before, I do not feel strongly about this, I was torn myself. On Thu, Oct 1, 2015 at 3:28 PM, Chesnay Scheplerwrote: > If we remove flink-staging because projects tend to get stuck there, what > mechanism prevents the same happening with flink-contrib? > > > On 01.10.2015 15:19, Stephan Ewen wrote: > >> +1 for Robert's comments. >> >> On Thu, Oct 1, 2015 at 3:16 PM, Robert Metzger >> wrote: >> >> Big +1 for graduating streaming out of staging. It is widely used, also in >>> production and we are spending a lot of effort into hardening it. >>> I also agree with the proposed new maven module structure. >>> >>> We have to carefully test the reworked structure for the scripts which >>> are >>> generating the hadoop1 and the scala 2.11 poms (they are transformed >>> using >>> a bunch of bash scripts). I can do that once the PR is open. >>> >>> @Chesnay: I would be fine with including the language binding into python >>> Where would new projects reside in, that previously would have been put >>> into flink-staging? >>> >>> flink-contrib >>> >>> >>> @Kostas: I understand the idea behind your suggested renaming, but thats >>> just a name. I don't think its going to influence how people are seeing >>> Flink: It doesn't feel like second class when adding >>> "flink-streaming-core" >>> to the dependencies to me. >>> Also, the "flink-datastream-scala" module would depend on >>> "flink-dataset-scala", which is kind of weird. >>> >>> >>> I'm wondering whether we should remove the "flink-test-utils" module. I >>> don't think its really necessary, because we can put the test jars into >>> the >>> flink-tests project and include them using the "test-jar" dependency. >>> >>> >>> On Thu, Oct 1, 2015 at 2:27 PM, Kostas Tzoumas >>> wrote: >>> >>> +1 I wanted to suggest that we rename modules to fully accept streaming as first class, qualifying also "batch" as "batch" (e.g., flink-java --> flink-dataset-java, flink-streaming --> flink-datastream, etc). This would break maven dependencies (temporary hell :-) so it's not a decision to take lightly. I'm not strongly advocating for it. On Thu, Oct 1, 2015 at 12:44 PM, Chesnay Schepler wrote: I like it in general. But while we're at it, what is the purpose of the > flink-tests project, or rather which tests belong there instead of the > individual projects? > > Where would new projects reside in, that previously would have been put > into flink-staging? > > Lastly, I'd like to merge flink-language-binding into flink-python. I > can >>> go more into detail but the gist of it is that the abstraction just > doesn't > work. > > > On 01.10.2015 12:40, Márton Balassi wrote: > > Great to see streaming graduating. :) >> >> I like the outline, both getting rid of staging, having the examples >> together and generally flattening the structure are very reasonable to >> > me. > You have listed flink-streaming-examples under >> > flink-streaming-connectors > and left out some less prominent maven modules, but I assume the first >> > is > accidental while the second is intentional to make the list a bit >> > briefer. > Best, >> >> Marton >> >> >> On Thu, Oct 1, 2015 at 12:25 PM, Stephan Ewen >> > wrote: >>> Hi all! >> >>> We are making good headway with reworking the last parts of the >>> >> Window >>> API. >>> After that, the streaming API should be good to be pulled out of >>> >> staging. > Since we are reorganizing the projects as part of that, I would >>> >> shift a >>> bit >>> more to bring things a bit more up to date. >>> >>> In this restructure, I would like to get rid of the "flink-staging" >>> project. Anyone who only uses the maven artifacts sees no difference >>> whether a project is in "staging" or not, so it does not help much to >>> have >>> that directory structure. >>> On the other hand, projects have a tendency to linger in staging >>> >> forever > (like avro, spargel, hbase, jdbc, ...) >>> >>> The new structure could be >>> >>> flink-core >>> flink-java >>> flink-scala >>> flink-streaming-core >>> flink-streaming-scala >>> >>> flink-runtime >>> flink-runtime-web >>> flink-optimizer >>> flink-clients >>> >>> flink-shaded >>> -> flink-shaded-hadoop >>> -> flink-shaded-hadoop2 >>> -> flink-shaded-include-yarn-tests >>> -> flink-shaded-curator >>> >>> flink-examples >>> -> (have all examples, Scala and Java, Batch and Streaming) >>> >>>
Re: Pulling Streaming out of staging and project restructure
+1 For pulling out and the restructure. Enough good arguments have been brought forward and I agree with all of them. On Thu, 1 Oct 2015 at 17:47 Ufuk Celebiwrote: > > > On 01 Oct 2015, at 16:48, Robert Metzger wrote: > > > > @Chesnay: Nothing prevents projects from getting stuck there. But at > least > > we don't have two staging repositories (dist, staging). Also I would not > > say that there has been no graduation out of staging. Yarn was also there > > once, streaming and gelly are currently leaving it. So I would actually > say > > it worked. > > > > One more thing: Can we actually put the web submission client into a > > separate module until we have something new and better? > > > > It does really not fit into "flink-clients". It makes Flink users pull > the > > jetty dependencies even if they never use them for example when running > > Flink from an IDE. > > I think most people use it as part of the binary distribution and nothing > else anyways. So +1.
Re: Pulling Streaming out of staging and project restructure
+1 for the new Maven project structure +1 for removing the flink-testing-utils module +1 for moving flink-language-binding to flink-python On Thu, Oct 1, 2015 at 6:27 PM, Aljoscha Krettekwrote: > +1 For pulling out and the restructure. Enough good arguments have been > brought forward and I agree with all of them. > > On Thu, 1 Oct 2015 at 17:47 Ufuk Celebi wrote: > >> >> > On 01 Oct 2015, at 16:48, Robert Metzger wrote: >> > >> > @Chesnay: Nothing prevents projects from getting stuck there. But at >> least >> > we don't have two staging repositories (dist, staging). Also I would not >> > say that there has been no graduation out of staging. Yarn was also there >> > once, streaming and gelly are currently leaving it. So I would actually >> say >> > it worked. >> > >> > One more thing: Can we actually put the web submission client into a >> > separate module until we have something new and better? >> > >> > It does really not fit into "flink-clients". It makes Flink users pull >> the >> > jetty dependencies even if they never use them for example when running >> > Flink from an IDE. >> >> I think most people use it as part of the binary distribution and nothing >> else anyways. So +1.
Hash-based aggregation
Hello, I would really like to see FLINK-2237 solved. I would implement this feature over the weekend, if the CompactingHashTable can be used to solve it (see my comment there). Could you please give me some advice on whether is this a viable approach, or you perhaps see some difficulties that I'm not aware of? Best, Gabor
Re: Pulling Streaming out of staging and project restructure
+1 I like the idea moving "staging" projects into appropriate modules. While we are at it, I would like to propose changing " flink-hadoop-compatibility" to "flink-hadoop". It is in my bucket list but would be nice if it is part of re-org. Supporting Hadoop in the connector implicitly means compatibility with Hadoop. Also same thing with "flink-storm-compatibility" to "flink-storm". - Henry On Thu, Oct 1, 2015 at 3:25 AM, Stephan Ewenwrote: > Hi all! > > We are making good headway with reworking the last parts of the Window API. > After that, the streaming API should be good to be pulled out of staging. > > Since we are reorganizing the projects as part of that, I would shift a bit > more to bring things a bit more up to date. > > In this restructure, I would like to get rid of the "flink-staging" > project. Anyone who only uses the maven artifacts sees no difference > whether a project is in "staging" or not, so it does not help much to have > that directory structure. > On the other hand, projects have a tendency to linger in staging forever > (like avro, spargel, hbase, jdbc, ...) > > The new structure could be > > flink-core > flink-java > flink-scala > flink-streaming-core > flink-streaming-scala > > flink-runtime > flink-runtime-web > flink-optimizer > flink-clients > > flink-shaded > -> flink-shaded-hadoop > -> flink-shaded-hadoop2 > -> flink-shaded-include-yarn-tests > -> flink-shaded-curator > > flink-examples > -> (have all examples, Scala and Java, Batch and Streaming) > > flink-batch-connectors > -> flink-avro > -> flink-jdbc > -> flink-hadoop-compatibility > -> flink-hbase > -> flink-hcatalog > > flink-streaming-connectors > -> flink-connector-twitter > -> flink-streaming-examples > -> flink-connector-flume > -> flink-connector-kafka > -> flink-connector-elasticsearch > -> flink-connector-rabbitmq > -> flink-connector-filesystem > > flink-libraries > -> flink-gelly > -> flink-gelly-scala > -> flink-ml > -> flink-table > -> flink-language-binding > -> flink-python > > > flink-scala-shell > > flink-test-utils > flink-tests > flink-fs-tests > > flink-contrib > -> flink-storm-compatibility > -> flink-storm-compatibility-examples > -> flink-streaming-utils > -> flink-tweet-inputformat > -> flink-operator-stats > -> flink-tez > > flink-quickstart > -> flink-quickstart-java > -> flink-quickstart-scala > -> flink-tez-quickstart > > flink-yarn > flink-yarn-tests > > flink-dist > > flink-benchmark > > > Let me know if that makes sense! > > Greetings, > Stephan
Re: Pulling Streaming out of staging and project restructure
I will commit something to flink-storm-compatibility tomorrow that contains some internal package restructuring. I think, renaming the three modules in this commit would be a smart move as both changes result in merge conflicts when rebasing open PRs. Thus we can limit this pain to a single time. If no objections, I will commit those changes tomorrow. -Matthias On 10/01/2015 09:52 PM, Henry Saputra wrote: > +1 > > I like the idea moving "staging" projects into appropriate modules. > > While we are at it, I would like to propose changing " > flink-hadoop-compatibility" to "flink-hadoop". It is in my bucket list > but would be nice if it is part of re-org. > Supporting Hadoop in the connector implicitly means compatibility with Hadoop. > Also same thing with "flink-storm-compatibility" to "flink-storm". > > - Henry > > On Thu, Oct 1, 2015 at 3:25 AM, Stephan Ewenwrote: >> Hi all! >> >> We are making good headway with reworking the last parts of the Window API. >> After that, the streaming API should be good to be pulled out of staging. >> >> Since we are reorganizing the projects as part of that, I would shift a bit >> more to bring things a bit more up to date. >> >> In this restructure, I would like to get rid of the "flink-staging" >> project. Anyone who only uses the maven artifacts sees no difference >> whether a project is in "staging" or not, so it does not help much to have >> that directory structure. >> On the other hand, projects have a tendency to linger in staging forever >> (like avro, spargel, hbase, jdbc, ...) >> >> The new structure could be >> >> flink-core >> flink-java >> flink-scala >> flink-streaming-core >> flink-streaming-scala >> >> flink-runtime >> flink-runtime-web >> flink-optimizer >> flink-clients >> >> flink-shaded >> -> flink-shaded-hadoop >> -> flink-shaded-hadoop2 >> -> flink-shaded-include-yarn-tests >> -> flink-shaded-curator >> >> flink-examples >> -> (have all examples, Scala and Java, Batch and Streaming) >> >> flink-batch-connectors >> -> flink-avro >> -> flink-jdbc >> -> flink-hadoop-compatibility >> -> flink-hbase >> -> flink-hcatalog >> >> flink-streaming-connectors >> -> flink-connector-twitter >> -> flink-streaming-examples >> -> flink-connector-flume >> -> flink-connector-kafka >> -> flink-connector-elasticsearch >> -> flink-connector-rabbitmq >> -> flink-connector-filesystem >> >> flink-libraries >> -> flink-gelly >> -> flink-gelly-scala >> -> flink-ml >> -> flink-table >> -> flink-language-binding >> -> flink-python >> >> >> flink-scala-shell >> >> flink-test-utils >> flink-tests >> flink-fs-tests >> >> flink-contrib >> -> flink-storm-compatibility >> -> flink-storm-compatibility-examples >> -> flink-streaming-utils >> -> flink-tweet-inputformat >> -> flink-operator-stats >> -> flink-tez >> >> flink-quickstart >> -> flink-quickstart-java >> -> flink-quickstart-scala >> -> flink-tez-quickstart >> >> flink-yarn >> flink-yarn-tests >> >> flink-dist >> >> flink-benchmark >> >> >> Let me know if that makes sense! >> >> Greetings, >> Stephan signature.asc Description: OpenPGP digital signature