Github user ottobackwards commented on the issue:
https://github.com/apache/incubator-metron/pull/467
There are two things here I *think*
1. The test assumptions and the failure on some platforms ( if this is
indeed the problem )
2. What this means for the real use of the
Github user ottobackwards commented on the issue:
https://github.com/apache/incubator-metron/pull/467
What are the performance penalties of doing this?
Part of the discussion in
[HADOOP-12009](https://issues.apache.org/jira/browse/HADOOP-12009) that lead to
the
Github user ottobackwards commented on the issue:
https://github.com/apache/incubator-metron/pull/466
[INFO]
[INFO] Reactor Summary:
[INFO]
[INFO] Metron .
METRON-743 (https://github.com/apache/incubator-metron/pull/467) for
reference.
On Sat, Feb 25, 2017 at 11:51 PM, Casey Stella wrote:
> Hmm, that's a very good catch if it's the issue. I was able to verify
> that if you botch the sort order of the files that it fails.
>
>
GitHub user cestella opened a pull request:
https://github.com/apache/incubator-metron/pull/467
METRON-743: Sort the files when reading results from Pcap
The FileSystem.listFiles() call does not return the files in sorted order,
which we assume for all FileSystem implementations.
Github user cestella commented on the issue:
https://github.com/apache/incubator-metron/pull/466
It shouldn't because the tests are already run with the code submitted in
this PR. This is just about getting git in sync with what's generated.
---
If your project is set up for it,
Github user ottobackwards commented on the issue:
https://github.com/apache/incubator-metron/pull/466
Does building or running the tests have anything to do with it?
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If
Hmm, that's a very good catch if it's the issue. I was able to verify that
if you botch the sort order of the files that it fails.
Would you mind sorting the files on PcapJob line 199 by filename?
Something like Collections.sort(files, (o1,o2) ->
o1.getName().compareTo(o2.getName()));
I'm going
Github user cestella commented on the issue:
https://github.com/apache/incubator-metron/pull/466
Yep
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if
Github user ottobackwards commented on the issue:
https://github.com/apache/incubator-metron/pull/466
So - the test is - build without changing anything after clone, and there
should not be shown in git status right?
---
If your project is set up for it, you can reply to this email
Well, if you are going to be so quick, how can I make you wait? working on
it.
On February 25, 2017 at 23:39:15, Casey Stella (ceste...@gmail.com) wrote:
Sorry, couldn't wait for monday. Fix is submitted for your review and
heckling: https://github.com/apache/incubator-metron/pull/466 :)
On
Sorry, couldn't wait for monday. Fix is submitted for your review and
heckling: https://github.com/apache/incubator-metron/pull/466 :)
On Sat, Feb 25, 2017 at 11:22 PM, Matt Foley wrote:
> Well, the last three are in a path with “/generated/” in it.
>
>
> On 2/25/17, 7:39 PM,
GitHub user cestella opened a pull request:
https://github.com/apache/incubator-metron/pull/466
METRON-742: Generated code for profile window selector DSL did not get
committed as part of METRON-690
Lest we be concerned this implies that the testing done on 690 is voided,
this was
/**
* List the statuses and block locations of the files in the given path.
* Does not guarantee to return the iterator that traverses statuses
* of the files in a sorted order.
*
* If the path is a directory,
* if recursive is false, returns files in the directory;
* if recursive is
Well, the last three are in a path with “/generated/” in it.
On 2/25/17, 7:39 PM, "Casey Stella" wrote:
Crap, those are generated by the new profile selector dsl committed Friday.
I must've missed them on a commit on the PR. They are generated by the
build, so
https://issues.apache.org/jira/browse/HADOOP-12009 makes it seem like
there is no order
On February 25, 2017 at 23:06:37, Otto Fowler (ottobackwa...@gmail.com)
wrote:
Maybe Hadoop Local FileSystem returns different things from ListFiles() on
different platforms?
That would be something to
Maybe Hadoop Local FileSystem returns different things from ListFiles() on
different platforms?
That would be something to check?
Sorry that is all I got right now
On February 25, 2017 at 22:57:49, Otto Fowler (ottobackwa...@gmail.com)
wrote:
There are also some if Log.isDebugEnabled()
There are multiple “tests” within the test, with different parameters. If
you look at where this is breaking, it is at
{
//make sure I get them all.
Iterable results =
job.query(new Path(outDir.getAbsolutePath())
, new Path(queryDir.getAbsolutePath())
Crap, those are generated by the new profile selector dsl committed Friday.
I must've missed them on a commit on the PR. They are generated by the
build, so it was factored into the tests and such for the PR. Sorry for the
inconvenience, I'll have to make another PR on Monday to get them in.
On
That's my guess. I noticed the same thing.
-Kyle
> On Feb 25, 2017, at 8:31 PM, Otto Fowler wrote:
>
> Changes not staged for commit:
>
> (use "git add ..." to update what will be committed)
>
> (use "git checkout -- ..." to discard changes in working directory)
>
mvn integration-test
Although I have also tried...
mvn clean install && mvn integration-test
mvn clean package && mvn integration-test
mvn install && mvn surefire-test@unit-tests && mvn
surefire-test@integration-tests
-Kyle
> On Feb 25, 2017, at 8:34 PM, Otto Fowler
What command are you using to build?
On February 25, 2017 at 17:40:20, Kyle Richardson (kylerichards...@gmail.com)
wrote:
Tried with Oracle JDK and got the same result. I went as far as trying to
run it through the debugger but am not that familiar with this part of the
code. The timestamps of
Changes not staged for commit:
(use "git add ..." to update what will be committed)
(use "git checkout -- ..." to discard changes in working directory)
modified:
metron-analytics/metron-profiler-client/src/main/java/Window.tokens
modified:
Tried with Oracle JDK and got the same result. I went as far as trying to run
it through the debugger but am not that familiar with this part of the code.
The timestamps of the packets are definitely not coming back in the expected
order, but I'm not sure why. Could it be related to something
16GB. It's a MBP from a few years back. I'm getting the same result as most
it seems, currently oversubscribed but doable with care, works way better
with the Ansible/Ambari integration that I'm doing for METRON-671.
-D...
On Sat, Feb 25, 2017 at 2:13 PM, Matt Foley wrote:
>
David, how much RAM do you have in the test system where you run full dev?
On 2/25/17, 8:46 AM, "David Lyle" wrote:
+1 (binding)
Checked signatures
Tests passed
Full dev ran sucessfully-ish (bit more below, tl;dr- it's ugly, but much
better after
+1 (binding)
Checked signatures
Tests passed
Full dev ran sucessfully-ish (bit more below, tl;dr- it's ugly, but much
better after METRON-671)
Data flowed into the index/dashboard/hdfs
On full-dev: Monit got into a state where it started killing topologies
because the Storm cli was non-responsive
I'm running it up now. I don't think it should affect the release, but it's
extremely important that full-dev work. It's how quick-dev is created. Any
notion that it's been deprecated is incorrect.
Fwiw, I've been working with it for the last few weeks as part of
METRON-671 and it's been
Of course the same commands build locally
On February 25, 2017 at 09:17:50, Otto Fowler (ottobackwa...@gmail.com)
wrote:
https://s3.amazonaws.com/archive.travis-ci.org/jobs/205276680/log.txt
On February 25, 2017 at 08:20:54, Casey Stella (ceste...@gmail.com) wrote:
I have not seen those "no
I see this with full dev on my laptop pretty much every time.
I think it is a resource issue. I see not enough memory errors trying to
start things.
On February 25, 2017 at 09:15:15, Ryan Merriman (merrim...@gmail.com) wrote:
When I go to Ambari to ensure the services are all up, HDFS is
When I go to Ambari to ensure the services are all up, HDFS is down. I tried
it 4 or 5 times and got the same result each time. I've seen others validate
with quick dev so I assumed full dev was not used anymore. I'll spin it up
this morning and get a more detailed error.
Is anyone else
Ok, I've tried this so many times I may be going crazy, so thought I'd ask
the community for a sanity check.
I'm trying to verify RC5 and I keep running into the same integration test
failures but only on my Fedora (24 and 25) and CentOS 7 systems. It passes
fine on my Macbook.
It always fails
What exactly are the errors that you saw, Ryan?
On Sat, Feb 25, 2017 at 07:31 David Lyle wrote:
> Is there any reason full dev shouldn't be working?
>
> On Fri, Feb 24, 2017 at 9:19 PM, Casey Stella wrote:
>
> > Sounds like a good idea to me; thanks
I have not seen those "no output received for 10m errors before. Can you
change the Travis command to not have -q for maven so we can see more
context?
On Sat, Feb 25, 2017 at 07:56 Otto Fowler wrote:
> I have not had a build work on Travis CI ( linked to my fork ) in 4
Is there any reason full dev shouldn't be working?
On Fri, Feb 24, 2017 at 9:19 PM, Casey Stella wrote:
> Sounds like a good idea to me; thanks Ryan!
> On Fri, Feb 24, 2017 at 21:11 Ryan Merriman wrote:
>
> > +1 binding
> >
> > Verified the signature
>
35 matches
Mail list logo