You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@storm.apache.org by Bipin Prasad <bi...@apache.org> on 2023/06/16 15:57:21 UTC

Storm Cassandra

All,
   I am working on creating a new release for Storm 2.5.0. Unfortunately,
storm-cassandra has been failing tests since we moved to Git Actions. Can
someone take a look at it and possibly get it to work - with a newer
version of Cassandra if possible and/or upgraded tests?

   If no one is using Storm Cassandra (under xternal/storm-cassandra), I
can exclude this from the build and release). Open to other suggestions.

Thanks
--Bipin

Re: Storm Cassandra

Posted by Alexandre Vermeerbergen <av...@gmail.com>.
+1 to exclude Storm Cassandra
(my 2 cents)

Thanks,
Alex

Le ven. 16 juin 2023 à 17:58, Bipin Prasad <bi...@apache.org> a écrit :
>
> All,
>    I am working on creating a new release for Storm 2.5.0. Unfortunately,
> storm-cassandra has been failing tests since we moved to Git Actions. Can
> someone take a look at it and possibly get it to work - with a newer
> version of Cassandra if possible and/or upgraded tests?
>
>    If no one is using Storm Cassandra (under xternal/storm-cassandra), I
> can exclude this from the build and release). Open to other suggestions.
>
> Thanks
> --Bipin

Re: Storm Cassandra

Posted by Richard Zowalla <rz...@apache.org>.
The PR by Bipin seems to fix it? ;)

Am Freitag, dem 16.06.2023 um 20:19 +0200 schrieb Richard Zowalla:
> Ok from looking into gitact-script.sh, it looks like we are skipping
> storm-cassandra, storm-hive, storm-hdfs, storm-hbase, storm-sql,
> storm-
> hdfs-blobstore if we are building with Java 11 and run them with Java
> 8. That explains why Java 11 runs do not fail ;-)
> 
> Nevertheless, we have runs, which are just working fine with Java 8.
> If
> we fail, we fail (at the moment) on pull requests with a JVM crash,
> see
> [1].
> 
> It seems, that we are already running with 3g memory, so it might not
> be the root problem, but could be related (due to GH action limits /
> overall consumption in the org).
> 
> On master, it seems to be fine (just happens on some PRs).
> 
> [1] https://gist.github.com/rzo1/5aa510022f27bdf3c8eda66089774b68
> 
> 
> Am Freitag, dem 16.06.2023 um 17:25 +0000 schrieb Bipin Prasad:
> >  Created Jira [STORM-3923] Cassandra module fails tests probably
> > OOM
> > - ASF JIRA . I think mvn test is using 3gb (inherited from main
> > pom.xml), but will add directly in its pom under Cassandra. Also
> > MAVEN_OPTS in git actions is set to 1.5gb. I wonder if that is
> > limiting the memory.
> > 
> > > 
> > > 
> > >  | 
> > [STORM-3923] Cassandra module fails tests probably OOM - ASF JIRA
> > 
> > 
> >  |
> > 
> >  |
> > 
> >  |
> > 
> > 
> > 
> > 
> >     On Friday, June 16, 2023 at 09:58:32 AM PDT, Richard Zowalla
> > <ri...@zowalla.com> wrote:  
> >  
> >  Hi,
> > 
> > looks like it only falls occasionally in Java 8 with an OOM issue /
> > JVM crash. Might be possible to just increase the JVM memory for
> > that
> > module. 
> > 
> > I remember from the Travis migration, that we had some heap
> > adjustments in the scripts. Wdyt?
> > 
> > Gruß
> > Richard 
> > 
> > Am 16. Juni 2023 17:57:21 MESZ schrieb Bipin Prasad
> > <bi...@apache.org>:
> > > All,
> > >   I am working on creating a new release for Storm 2.5.0.
> > > Unfortunately,
> > > storm-cassandra has been failing tests since we moved to Git
> > > Actions. Can
> > > someone take a look at it and possibly get it to work - with a
> > > newer
> > > version of Cassandra if possible and/or upgraded tests?
> > > 
> > >   If no one is using Storm Cassandra (under xternal/storm-
> > > cassandra), I
> > > can exclude this from the build and release). Open to other
> > > suggestions.
> > > 
> > > Thanks
> > > --Bipin
> >   
> 


Re: Storm Cassandra

Posted by Richard Zowalla <ri...@zowalla.com>.
Ok from looking into gitact-script.sh, it looks like we are skipping
storm-cassandra, storm-hive, storm-hdfs, storm-hbase, storm-sql, storm-
hdfs-blobstore if we are building with Java 11 and run them with Java
8. That explains why Java 11 runs do not fail ;-)

Nevertheless, we have runs, which are just working fine with Java 8. If
we fail, we fail (at the moment) on pull requests with a JVM crash, see
[1].

It seems, that we are already running with 3g memory, so it might not
be the root problem, but could be related (due to GH action limits /
overall consumption in the org).

On master, it seems to be fine (just happens on some PRs).

[1] https://gist.github.com/rzo1/5aa510022f27bdf3c8eda66089774b68


Am Freitag, dem 16.06.2023 um 17:25 +0000 schrieb Bipin Prasad:
>  Created Jira [STORM-3923] Cassandra module fails tests probably OOM
> - ASF JIRA . I think mvn test is using 3gb (inherited from main
> pom.xml), but will add directly in its pom under Cassandra. Also
> MAVEN_OPTS in git actions is set to 1.5gb. I wonder if that is
> limiting the memory.
> 
> > 
> > 
> >  | 
> [STORM-3923] Cassandra module fails tests probably OOM - ASF JIRA
> 
> 
>  |
> 
>  |
> 
>  |
> 
> 
> 
> 
>     On Friday, June 16, 2023 at 09:58:32 AM PDT, Richard Zowalla
> <ri...@zowalla.com> wrote:  
>  
>  Hi,
> 
> looks like it only falls occasionally in Java 8 with an OOM issue /
> JVM crash. Might be possible to just increase the JVM memory for that
> module. 
> 
> I remember from the Travis migration, that we had some heap
> adjustments in the scripts. Wdyt?
> 
> Gruß
> Richard 
> 
> Am 16. Juni 2023 17:57:21 MESZ schrieb Bipin Prasad
> <bi...@apache.org>:
> > All,
> >   I am working on creating a new release for Storm 2.5.0.
> > Unfortunately,
> > storm-cassandra has been failing tests since we moved to Git
> > Actions. Can
> > someone take a look at it and possibly get it to work - with a
> > newer
> > version of Cassandra if possible and/or upgraded tests?
> > 
> >   If no one is using Storm Cassandra (under xternal/storm-
> > cassandra), I
> > can exclude this from the build and release). Open to other
> > suggestions.
> > 
> > Thanks
> > --Bipin
>   


Re: Storm Cassandra

Posted by Bipin Prasad <bi...@yahoo.com.INVALID>.
 Created Jira [STORM-3923] Cassandra module fails tests probably OOM - ASF JIRA . I think mvn test is using 3gb (inherited from main pom.xml), but will add directly in its pom under Cassandra. Also MAVEN_OPTS in git actions is set to 1.5gb. I wonder if that is limiting the memory.

| 
| 
|  | 
[STORM-3923] Cassandra module fails tests probably OOM - ASF JIRA


 |

 |

 |




    On Friday, June 16, 2023 at 09:58:32 AM PDT, Richard Zowalla <ri...@zowalla.com> wrote:  
 
 Hi,

looks like it only falls occasionally in Java 8 with an OOM issue / JVM crash. Might be possible to just increase the JVM memory for that module. 

I remember from the Travis migration, that we had some heap adjustments in the scripts. Wdyt?

Gruß
Richard 

Am 16. Juni 2023 17:57:21 MESZ schrieb Bipin Prasad <bi...@apache.org>:
>All,
>  I am working on creating a new release for Storm 2.5.0. Unfortunately,
>storm-cassandra has been failing tests since we moved to Git Actions. Can
>someone take a look at it and possibly get it to work - with a newer
>version of Cassandra if possible and/or upgraded tests?
>
>  If no one is using Storm Cassandra (under xternal/storm-cassandra), I
>can exclude this from the build and release). Open to other suggestions.
>
>Thanks
>--Bipin
  

Re: Storm Cassandra

Posted by Richard Zowalla <ri...@zowalla.com>.
Hi,

looks like it only falls occasionally in Java 8 with an OOM issue / JVM crash. Might be possible to just increase the JVM memory for that module. 

I remember from the Travis migration, that we had some heap adjustments in the scripts. Wdyt?

Gruß
Richard 

Am 16. Juni 2023 17:57:21 MESZ schrieb Bipin Prasad <bi...@apache.org>:
>All,
>   I am working on creating a new release for Storm 2.5.0. Unfortunately,
>storm-cassandra has been failing tests since we moved to Git Actions. Can
>someone take a look at it and possibly get it to work - with a newer
>version of Cassandra if possible and/or upgraded tests?
>
>   If no one is using Storm Cassandra (under xternal/storm-cassandra), I
>can exclude this from the build and release). Open to other suggestions.
>
>Thanks
>--Bipin