Re: trunk build does not work // was: mountByFS does not work

2016-11-30 Thread Sandro Boehme

Hi Robert,

Am 28.11.16 um 23:32 schrieb Robert Munteanu:

On Fri, 2016-11-18 at 00:22 +0100, Sandro Boehme wrote:

Hi Julian,

I like the idea of individual Jenkins jobs for the reasons you
pointed out!
If we would have a "nightly" build (whatever time that would be) we
could link it at the Getting and building Sling page. We could write
there that in case the build does currently not work (e.g. because
there
is some bigger work in progress or a plain oversight) they could have
a
look at what revision actually does work and use that.


I've updated the building Sling docs to reflect how we expect things to
work now, see especially

  http://sling.apache.org/documentation/development/getting-and-buildin
g-sling.html#tldr-short-form-build-run-instructions

  http://sling.apache.org/documentation/development/getting-and-buildin
g-sling.html#with-the-maven-command-line-tool

Hopefully that makes more sense now.


It absolutely does!
I've just had cases where 'mvn clean install sling:install' of a 
snapshot bundle to the normal launchpad seem to make problems with 
resolving bundles. But if it should generally work it's all good.


Best,

Sandro


Re: trunk build does not work // was: mountByFS does not work

2016-11-28 Thread Robert Munteanu
On Fri, 2016-11-18 at 00:22 +0100, Sandro Boehme wrote:
> Hi Julian,
> 
> I like the idea of individual Jenkins jobs for the reasons you
> pointed out!
> If we would have a "nightly" build (whatever time that would be) we 
> could link it at the Getting and building Sling page. We could write 
> there that in case the build does currently not work (e.g. because
> there 
> is some bigger work in progress or a plain oversight) they could have
> a 
> look at what revision actually does work and use that.

I've updated the building Sling docs to reflect how we expect things to
work now, see especially

  http://sling.apache.org/documentation/development/getting-and-buildin
g-sling.html#tldr-short-form-build-run-instructions

  http://sling.apache.org/documentation/development/getting-and-buildin
g-sling.html#with-the-maven-command-line-tool

Hopefully that makes more sense now.

Robert

> People who would stop trying Sling because of a non working build
> and 
> who would not ask at the mailing list would then have a chance to
> get 
> the build working anyways.
> WDYT?
> 
> Best,
> 
> Sandro
> 
> Am 17.11.16 um 14:53 schrieb Julian Sedding:
> > Hi Sandro
> > 
> > You are right. However, the full build on Jenkins failed too often.
> > Partly due to issues or flaky tests in Sling, partly due to e.g.
> > infrastructure issues. This made the Jenkins builds almost useless,
> > because chances were that it weren't your changes that made the
> > build
> > fail.
> > 
> > Therefore Robert added configurations to run each module
> > independently
> > on Jenkins. That gives us a better overview of which modules are ok
> > and which ones aren't. However, we lost the full build in the
> > process.
> > Maybe we could add it back and run it once every night. WDYT,
> > Robert?
> > 
> > Regards
> > Julian
> > 
> > 
> > 
> > On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme  > de> wrote:
> > > > According to our Jenkins job
> > > > 
> > > >   https://builds.apache.org/job/sling-launchpad-builder-1.8/
> > > > 
> > > > the launchpad should work without issues.
> > > > 
> > > > A couple of additional questions:
> > > > 
> > > > - what Java version do you use for building and running Sling?
> > > > - what happens if you try to build the project with "-U", to
> > > > force
> > > > updating the snapshots? The command would then be mvn -U clean
> > > > verify
> > > > 
> > > > Robert
> > > 
> > > According to the "Getting and Building Sling" page [1] I just did
> > > "svn co
> > > ...", "mvn clean install" (-DskipTests) and "java -jar ..." and
> > > expected to
> > > be able to build and use Sling. But if we don't have a Jenkins
> > > job that does
> > > the same a committer will probably more easily overlook if he
> > > breaks the
> > > build for people who are trying out Sling. But I guess you know
> > > that and I
> > > have the feeling that I miss something here.
> > > 
> > > I use Java 8. I didn't try -U but Julians changes fixed the
> > > problem for me.
> > > 
> > > 
> > > [1] -
> > > http://sling.apache.org/documentation/development/getting-and-bui
> > > lding-sling.html
> > > 
> > > Best,
> > > 
> > > Sandro
> 
> 



Re: trunk build does not work // was: mountByFS does not work

2016-11-28 Thread Bertrand Delacretaz
Hi Sandro,

On Sun, Nov 27, 2016 at 11:39 PM, Sandro Boehme  wrote:
> ...Is there a launchpad version somewhere that compiles the latest
> snapshots to the standalone jar?...

We've been discussing generating that [1] along with a "stable"
launchpad but that's not done yet.

-Bertrand

[1] 
https://cwiki.apache.org/confluence/display/SLING/Moving+the+Sling+Launchpad+to+use+released+artifacts+only


Re: trunk build does not work // was: mountByFS does not work

2016-11-27 Thread Sandro Boehme

Am 19.11.16 um 21:56 schrieb Sandro Boehme:

Am 17.11.16 um 23:22 schrieb Robert Munteanu:

On Thu, 2016-11-17 at 14:53 +0100, Julian Sedding wrote:

Hi Sandro

You are right. However, the full build on Jenkins failed too often.
Partly due to issues or flaky tests in Sling, partly due to e.g.
infrastructure issues. This made the Jenkins builds almost useless,
because chances were that it weren't your changes that made the build
fail.

Therefore Robert added configurations to run each module
independently
on Jenkins. That gives us a better overview of which modules are ok
and which ones aren't. However, we lost the full build in the
process.
Maybe we could add it back and run it once every night. WDYT, Robert?


First of all, I think that I should update the documentation :-) With
all the SNAPSHOTs being deployed by the Jenkins jobs the only command
that should be used to build Sling after the SVN checkout is

$ cd launchpad/builder
$ mvn -U clean install

which matches to a large degree what the Jenkins jobs do. There should
be absolutely no need to perform a full reactor build - which takes
time, and is not always stable - to get a launchpad.

Then I'd like to understand exactly why the reactor build fails, as it
should not, at least not when the individual module builds are working.

Robert

Hi Robert,

thanks a lot for the hint! Your magic commands make the build of the
server very quick. That makes my work with Sling a lot easier and that
would be true for everybody who reads it at the Sling website.

I guess that if I change a Sling Maven project and 'mvn install' it the
generated artifact is supposed to be picked up at Maven install phase of
the builder.

Best,

Sandro





Regards
Julian



On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme 

Re: trunk build does not work // was: mountByFS does not work

2016-11-19 Thread Sandro Boehme

Am 17.11.16 um 23:22 schrieb Robert Munteanu:

On Thu, 2016-11-17 at 14:53 +0100, Julian Sedding wrote:

Hi Sandro

You are right. However, the full build on Jenkins failed too often.
Partly due to issues or flaky tests in Sling, partly due to e.g.
infrastructure issues. This made the Jenkins builds almost useless,
because chances were that it weren't your changes that made the build
fail.

Therefore Robert added configurations to run each module
independently
on Jenkins. That gives us a better overview of which modules are ok
and which ones aren't. However, we lost the full build in the
process.
Maybe we could add it back and run it once every night. WDYT, Robert?


First of all, I think that I should update the documentation :-) With
all the SNAPSHOTs being deployed by the Jenkins jobs the only command
that should be used to build Sling after the SVN checkout is

$ cd launchpad/builder
$ mvn -U clean install

which matches to a large degree what the Jenkins jobs do. There should
be absolutely no need to perform a full reactor build - which takes
time, and is not always stable - to get a launchpad.

Then I'd like to understand exactly why the reactor build fails, as it
should not, at least not when the individual module builds are working.

Robert

Hi Robert,

thanks a lot for the hint! Your magic commands make the build of the 
server very quick. That makes my work with Sling a lot easier and that 
would be true for everybody who reads it at the Sling website.


I guess that if I change a Sling Maven project and 'mvn install' it the 
generated artifact is supposed to be picked up at Maven install phase of 
the builder.


Best,

Sandro





Regards
Julian



On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme 

Re: trunk build does not work // was: mountByFS does not work

2016-11-18 Thread Sandro Boehme

Am 17.11.16 um 23:22 schrieb Robert Munteanu:

On Thu, 2016-11-17 at 14:53 +0100, Julian Sedding wrote:

Hi Sandro

You are right. However, the full build on Jenkins failed too often.
Partly due to issues or flaky tests in Sling, partly due to e.g.
infrastructure issues. This made the Jenkins builds almost useless,
because chances were that it weren't your changes that made the build
fail.

Therefore Robert added configurations to run each module
independently
on Jenkins. That gives us a better overview of which modules are ok
and which ones aren't. However, we lost the full build in the
process.
Maybe we could add it back and run it once every night. WDYT, Robert?


First of all, I think that I should update the documentation :-) With
all the SNAPSHOTs being deployed by the Jenkins jobs the only command
that should be used to build Sling after the SVN checkout is

$ cd launchpad/builder
$ mvn -U clean install

which matches to a large degree what the Jenkins jobs do. There should
be absolutely no need to perform a full reactor build - which takes
time, and is not always stable - to get a launchpad.

Then I'd like to understand exactly why the reactor build fails, as it
should not, at least not when the individual module builds are working.

Robert

I don't know exactly why the reactor build failed.

In the next days I will have a look on that to find out what's the 
difference between the reactor build and the launchpad build. But of 
course a quick and stable build sounds very very good to me!


Best,

Sandro






Regards
Julian



On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme 

Re: trunk build does not work // was: mountByFS does not work

2016-11-17 Thread Sandro Boehme

Hi Julian,

I like the idea of individual Jenkins jobs for the reasons you pointed out!
If we would have a "nightly" build (whatever time that would be) we 
could link it at the Getting and building Sling page. We could write 
there that in case the build does currently not work (e.g. because there 
is some bigger work in progress or a plain oversight) they could have a 
look at what revision actually does work and use that.
People who would stop trying Sling because of a non working build and 
who would not ask at the mailing list would then have a chance to get 
the build working anyways.

WDYT?

Best,

Sandro

Am 17.11.16 um 14:53 schrieb Julian Sedding:

Hi Sandro

You are right. However, the full build on Jenkins failed too often.
Partly due to issues or flaky tests in Sling, partly due to e.g.
infrastructure issues. This made the Jenkins builds almost useless,
because chances were that it weren't your changes that made the build
fail.

Therefore Robert added configurations to run each module independently
on Jenkins. That gives us a better overview of which modules are ok
and which ones aren't. However, we lost the full build in the process.
Maybe we could add it back and run it once every night. WDYT, Robert?

Regards
Julian



On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme  wrote:

According to our Jenkins job

  https://builds.apache.org/job/sling-launchpad-builder-1.8/

the launchpad should work without issues.

A couple of additional questions:

- what Java version do you use for building and running Sling?
- what happens if you try to build the project with "-U", to force
updating the snapshots? The command would then be mvn -U clean verify

Robert


According to the "Getting and Building Sling" page [1] I just did "svn co
...", "mvn clean install" (-DskipTests) and "java -jar ..." and expected to
be able to build and use Sling. But if we don't have a Jenkins job that does
the same a committer will probably more easily overlook if he breaks the
build for people who are trying out Sling. But I guess you know that and I
have the feeling that I miss something here.

I use Java 8. I didn't try -U but Julians changes fixed the problem for me.


[1] -
http://sling.apache.org/documentation/development/getting-and-building-sling.html

Best,

Sandro






Re: trunk build does not work // was: mountByFS does not work

2016-11-17 Thread Robert Munteanu
On Thu, 2016-11-17 at 14:53 +0100, Julian Sedding wrote:
> Hi Sandro
> 
> You are right. However, the full build on Jenkins failed too often.
> Partly due to issues or flaky tests in Sling, partly due to e.g.
> infrastructure issues. This made the Jenkins builds almost useless,
> because chances were that it weren't your changes that made the build
> fail.
> 
> Therefore Robert added configurations to run each module
> independently
> on Jenkins. That gives us a better overview of which modules are ok
> and which ones aren't. However, we lost the full build in the
> process.
> Maybe we could add it back and run it once every night. WDYT, Robert?

First of all, I think that I should update the documentation :-) With
all the SNAPSHOTs being deployed by the Jenkins jobs the only command
that should be used to build Sling after the SVN checkout is

$ cd launchpad/builder
$ mvn -U clean install

which matches to a large degree what the Jenkins jobs do. There should
be absolutely no need to perform a full reactor build - which takes
time, and is not always stable - to get a launchpad.

Then I'd like to understand exactly why the reactor build fails, as it
should not, at least not when the individual module builds are working.

Robert

> 
> Regards
> Julian
> 
> 
> 
> On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme  > wrote:
> > > According to our Jenkins job
> > > 
> > >   https://builds.apache.org/job/sling-launchpad-builder-1.8/
> > > 
> > > the launchpad should work without issues.
> > > 
> > > A couple of additional questions:
> > > 
> > > - what Java version do you use for building and running Sling?
> > > - what happens if you try to build the project with "-U", to
> > > force
> > > updating the snapshots? The command would then be mvn -U clean
> > > verify
> > > 
> > > Robert
> > 
> > According to the "Getting and Building Sling" page [1] I just did
> > "svn co
> > ...", "mvn clean install" (-DskipTests) and "java -jar ..." and
> > expected to
> > be able to build and use Sling. But if we don't have a Jenkins job
> > that does
> > the same a committer will probably more easily overlook if he
> > breaks the
> > build for people who are trying out Sling. But I guess you know
> > that and I
> > have the feeling that I miss something here.
> > 
> > I use Java 8. I didn't try -U but Julians changes fixed the problem
> > for me.
> > 
> > 
> > [1] -
> > http://sling.apache.org/documentation/development/getting-and-build
> > ing-sling.html
> > 
> > Best,
> > 
> > Sandro



Re: trunk build does not work // was: mountByFS does not work

2016-11-17 Thread Julian Sedding
Hi Sandro

You are right. However, the full build on Jenkins failed too often.
Partly due to issues or flaky tests in Sling, partly due to e.g.
infrastructure issues. This made the Jenkins builds almost useless,
because chances were that it weren't your changes that made the build
fail.

Therefore Robert added configurations to run each module independently
on Jenkins. That gives us a better overview of which modules are ok
and which ones aren't. However, we lost the full build in the process.
Maybe we could add it back and run it once every night. WDYT, Robert?

Regards
Julian



On Wed, Nov 16, 2016 at 10:02 PM, Sandro Boehme  wrote:
>> According to our Jenkins job
>>
>>   https://builds.apache.org/job/sling-launchpad-builder-1.8/
>>
>> the launchpad should work without issues.
>>
>> A couple of additional questions:
>>
>> - what Java version do you use for building and running Sling?
>> - what happens if you try to build the project with "-U", to force
>> updating the snapshots? The command would then be mvn -U clean verify
>>
>> Robert
>
> According to the "Getting and Building Sling" page [1] I just did "svn co
> ...", "mvn clean install" (-DskipTests) and "java -jar ..." and expected to
> be able to build and use Sling. But if we don't have a Jenkins job that does
> the same a committer will probably more easily overlook if he breaks the
> build for people who are trying out Sling. But I guess you know that and I
> have the feeling that I miss something here.
>
> I use Java 8. I didn't try -U but Julians changes fixed the problem for me.
>
>
> [1] -
> http://sling.apache.org/documentation/development/getting-and-building-sling.html
>
> Best,
>
> Sandro


Re: trunk build does not work // was: mountByFS does not work

2016-11-16 Thread Sandro Boehme

According to our Jenkins job

  https://builds.apache.org/job/sling-launchpad-builder-1.8/

the launchpad should work without issues.

A couple of additional questions:

- what Java version do you use for building and running Sling?
- what happens if you try to build the project with "-U", to force
updating the snapshots? The command would then be mvn -U clean verify

Robert
According to the "Getting and Building Sling" page [1] I just did "svn 
co ...", "mvn clean install" (-DskipTests) and "java -jar ..." and 
expected to be able to build and use Sling. But if we don't have a 
Jenkins job that does the same a committer will probably more easily 
overlook if he breaks the build for people who are trying out Sling. But 
I guess you know that and I have the feeling that I miss something here.


I use Java 8. I didn't try -U but Julians changes fixed the problem for me.


[1] - 
http://sling.apache.org/documentation/development/getting-and-building-sling.html


Best,

Sandro


Re: trunk build does not work // was: mountByFS does not work

2016-11-16 Thread Sandro Boehme

Hi Julian,

thanks for the changes! It works for me as well now.

Best,

Sandro

Am 16.11.16 um 13:29 schrieb Julian Sedding:

Hi Sandro

I have committed my changes to these modules and run the full build
multiple times. So I expect that you should be able to run a
successful trunk build now.

Regards
Julian


On Wed, Nov 16, 2016 at 1:18 PM, Robert Munteanu  wrote:

On Tue, 2016-11-15 at 15:51 +0100, Sandro Boehme wrote:

Hi Julian,

thanks for your feedback!

Am 15.11.16 um 10:47 schrieb Julian Sedding:

Hi Sandro

org/apache/sling/jcr/base/LoginAdminWhitelist looks like your
checkout
or snapshots on your machine are not quite up to date. I removed
the


I'am at the latest trunk, did 'mvn clean ...' and created a new
sling
home with the '-c' option at server startup. What else do I have to
do
to be up to date?


method AbstractSlingRepositoryManager.getLoginAdminWhitelist(), so
maybe you need to first build a local snapshot of bundles/jcr/base
and
then try building bundles/jcr/oak-server again.


What version of jcr/base is compatible with which version of jcr/oak-
server?



Sorry for the inconvenience. I should soon be done with refactoring
these modules.


Does that mean there is nothing I can do right now?


According to our Jenkins job

  https://builds.apache.org/job/sling-launchpad-builder-1.8/

the launchpad should work without issues.

A couple of additional questions:

- what Java version do you use for building and running Sling?
- what happens if you try to build the project with "-U", to force
updating the snapshots? The command would then be mvn -U clean verify

Robert



Best,

Sandro



Regards
Julian


On Mon, Nov 14, 2016 at 9:50 PM, Sandro Boehme  wrote:

Hi Robert,


It took me a while to get the trunk to build. I had to
disable the
Rat
checks and update the dependency to org.apache.sling.jcr.base
version
2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
Why isn't there a build job anymore for trunk? Should I build
Sling
differently than described at the website [1]?



The context as is at [2]. Basically, the big 'trunk' jobs were
too slow
and flaky. Now we have many small jobs, with none of them slow
and a
couple of them flaky :-)

But the reactor build should work nonetheless.

I currently can't build it due to repository.apache.org
timeouts, can
you post any errors that you get?

Thanks,

Robert


[2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Je
nkins+Se
tup



The dependency to oasling.jcr.base 2.4.x has been changed by
Julian in the
meantime to prepare for the oak.server-1.1.2 release.
So I could build the trunk a few minutes ago.
But the server doesn't seem to start completely. When accessing
http://localhost:8080 I get an HTTP 503 saying
"Problem accessing /. Reason:
AuthenticationSupport service missing. Cannot authenticate
request."
138 bundles are available and active.

The error log shows me this error:
14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManage
r start:
Uncaught Throwable trying to access Repository, calling
stopRepository()
java.lang.AbstractMethodError:
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLogin
AdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
at
org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministr
ative(AbstractSlingRepository2.java:373)
at
org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loade
r.java:326)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerNamespac
es(Loader.java:213)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerBundleIn
ternal(Loader.java:225)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(L
oader.java:158)
at
org.apache.sling.jcr.base.internal.loader.Loader.(Loader.ja
va:75)
at
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(Ab
stractSlingRepositoryManager.java:371)
...

Any advice would be much appreciated!

Best,
Sandro











Re: trunk build does not work // was: mountByFS does not work

2016-11-16 Thread Julian Sedding
Hi Sandro

I have committed my changes to these modules and run the full build
multiple times. So I expect that you should be able to run a
successful trunk build now.

Regards
Julian


On Wed, Nov 16, 2016 at 1:18 PM, Robert Munteanu  wrote:
> On Tue, 2016-11-15 at 15:51 +0100, Sandro Boehme wrote:
>> Hi Julian,
>>
>> thanks for your feedback!
>>
>> Am 15.11.16 um 10:47 schrieb Julian Sedding:
>> > Hi Sandro
>> >
>> > org/apache/sling/jcr/base/LoginAdminWhitelist looks like your
>> > checkout
>> > or snapshots on your machine are not quite up to date. I removed
>> > the
>>
>> I'am at the latest trunk, did 'mvn clean ...' and created a new
>> sling
>> home with the '-c' option at server startup. What else do I have to
>> do
>> to be up to date?
>>
>> > method AbstractSlingRepositoryManager.getLoginAdminWhitelist(), so
>> > maybe you need to first build a local snapshot of bundles/jcr/base
>> > and
>> > then try building bundles/jcr/oak-server again.
>>
>> What version of jcr/base is compatible with which version of jcr/oak-
>> server?
>>
>> >
>> > Sorry for the inconvenience. I should soon be done with refactoring
>> > these modules.
>>
>> Does that mean there is nothing I can do right now?
>
> According to our Jenkins job
>
>   https://builds.apache.org/job/sling-launchpad-builder-1.8/
>
> the launchpad should work without issues.
>
> A couple of additional questions:
>
> - what Java version do you use for building and running Sling?
> - what happens if you try to build the project with "-U", to force
> updating the snapshots? The command would then be mvn -U clean verify
>
> Robert
>
>>
>> Best,
>>
>> Sandro
>>
>> >
>> > Regards
>> > Julian
>> >
>> >
>> > On Mon, Nov 14, 2016 at 9:50 PM, Sandro Boehme > > e> wrote:
>> > > Hi Robert,
>> > >
>> > > > > It took me a while to get the trunk to build. I had to
>> > > > > disable the
>> > > > > Rat
>> > > > > checks and update the dependency to org.apache.sling.jcr.base
>> > > > > version
>> > > > > 2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
>> > > > > Why isn't there a build job anymore for trunk? Should I build
>> > > > > Sling
>> > > > > differently than described at the website [1]?
>> > > >
>> > > >
>> > > > The context as is at [2]. Basically, the big 'trunk' jobs were
>> > > > too slow
>> > > > and flaky. Now we have many small jobs, with none of them slow
>> > > > and a
>> > > > couple of them flaky :-)
>> > > >
>> > > > But the reactor build should work nonetheless.
>> > > >
>> > > > I currently can't build it due to repository.apache.org
>> > > > timeouts, can
>> > > > you post any errors that you get?
>> > > >
>> > > > Thanks,
>> > > >
>> > > > Robert
>> > > >
>> > > >
>> > > > [2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Je
>> > > > nkins+Se
>> > > > tup
>> > >
>> > >
>> > > The dependency to oasling.jcr.base 2.4.x has been changed by
>> > > Julian in the
>> > > meantime to prepare for the oak.server-1.1.2 release.
>> > > So I could build the trunk a few minutes ago.
>> > > But the server doesn't seem to start completely. When accessing
>> > > http://localhost:8080 I get an HTTP 503 saying
>> > > "Problem accessing /. Reason:
>> > > AuthenticationSupport service missing. Cannot authenticate
>> > > request."
>> > > 138 bundles are available and active.
>> > >
>> > > The error log shows me this error:
>> > > 14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
>> > > org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManage
>> > > r start:
>> > > Uncaught Throwable trying to access Repository, calling
>> > > stopRepository()
>> > > java.lang.AbstractMethodError:
>> > > org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLogin
>> > > AdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
>> > > at
>> > > org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministr
>> > > ative(AbstractSlingRepository2.java:373)
>> > > at
>> > > org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loade
>> > > r.java:326)
>> > > at
>> > > org.apache.sling.jcr.base.internal.loader.Loader.registerNamespac
>> > > es(Loader.java:213)
>> > > at
>> > > org.apache.sling.jcr.base.internal.loader.Loader.registerBundleIn
>> > > ternal(Loader.java:225)
>> > > at
>> > > org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(L
>> > > oader.java:158)
>> > > at
>> > > org.apache.sling.jcr.base.internal.loader.Loader.(Loader.ja
>> > > va:75)
>> > > at
>> > > org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(Ab
>> > > stractSlingRepositoryManager.java:371)
>> > > ...
>> > >
>> > > Any advice would be much appreciated!
>> > >
>> > > Best,
>> > > Sandro
>>
>>
>


Re: trunk build does not work // was: mountByFS does not work

2016-11-16 Thread Robert Munteanu
On Tue, 2016-11-15 at 15:51 +0100, Sandro Boehme wrote:
> Hi Julian,
> 
> thanks for your feedback!
> 
> Am 15.11.16 um 10:47 schrieb Julian Sedding:
> > Hi Sandro
> > 
> > org/apache/sling/jcr/base/LoginAdminWhitelist looks like your
> > checkout
> > or snapshots on your machine are not quite up to date. I removed
> > the
> 
> I'am at the latest trunk, did 'mvn clean ...' and created a new
> sling 
> home with the '-c' option at server startup. What else do I have to
> do 
> to be up to date?
> 
> > method AbstractSlingRepositoryManager.getLoginAdminWhitelist(), so
> > maybe you need to first build a local snapshot of bundles/jcr/base
> > and
> > then try building bundles/jcr/oak-server again.
> 
> What version of jcr/base is compatible with which version of jcr/oak-
> server?
> 
> > 
> > Sorry for the inconvenience. I should soon be done with refactoring
> > these modules.
> 
> Does that mean there is nothing I can do right now?

According to our Jenkins job 

  https://builds.apache.org/job/sling-launchpad-builder-1.8/

the launchpad should work without issues.

A couple of additional questions:

- what Java version do you use for building and running Sling?
- what happens if you try to build the project with "-U", to force
updating the snapshots? The command would then be mvn -U clean verify

Robert

> 
> Best,
> 
> Sandro
> 
> > 
> > Regards
> > Julian
> > 
> > 
> > On Mon, Nov 14, 2016 at 9:50 PM, Sandro Boehme  > e> wrote:
> > > Hi Robert,
> > > 
> > > > > It took me a while to get the trunk to build. I had to
> > > > > disable the
> > > > > Rat
> > > > > checks and update the dependency to org.apache.sling.jcr.base
> > > > > version
> > > > > 2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
> > > > > Why isn't there a build job anymore for trunk? Should I build
> > > > > Sling
> > > > > differently than described at the website [1]?
> > > > 
> > > > 
> > > > The context as is at [2]. Basically, the big 'trunk' jobs were
> > > > too slow
> > > > and flaky. Now we have many small jobs, with none of them slow
> > > > and a
> > > > couple of them flaky :-)
> > > > 
> > > > But the reactor build should work nonetheless.
> > > > 
> > > > I currently can't build it due to repository.apache.org
> > > > timeouts, can
> > > > you post any errors that you get?
> > > > 
> > > > Thanks,
> > > > 
> > > > Robert
> > > > 
> > > > 
> > > > [2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Je
> > > > nkins+Se
> > > > tup
> > > 
> > > 
> > > The dependency to oasling.jcr.base 2.4.x has been changed by
> > > Julian in the
> > > meantime to prepare for the oak.server-1.1.2 release.
> > > So I could build the trunk a few minutes ago.
> > > But the server doesn't seem to start completely. When accessing
> > > http://localhost:8080 I get an HTTP 503 saying
> > > "Problem accessing /. Reason:
> > > AuthenticationSupport service missing. Cannot authenticate
> > > request."
> > > 138 bundles are available and active.
> > > 
> > > The error log shows me this error:
> > > 14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
> > > org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManage
> > > r start:
> > > Uncaught Throwable trying to access Repository, calling
> > > stopRepository()
> > > java.lang.AbstractMethodError:
> > > org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLogin
> > > AdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
> > > at
> > > org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministr
> > > ative(AbstractSlingRepository2.java:373)
> > > at
> > > org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loade
> > > r.java:326)
> > > at
> > > org.apache.sling.jcr.base.internal.loader.Loader.registerNamespac
> > > es(Loader.java:213)
> > > at
> > > org.apache.sling.jcr.base.internal.loader.Loader.registerBundleIn
> > > ternal(Loader.java:225)
> > > at
> > > org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(L
> > > oader.java:158)
> > > at
> > > org.apache.sling.jcr.base.internal.loader.Loader.(Loader.ja
> > > va:75)
> > > at
> > > org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(Ab
> > > stractSlingRepositoryManager.java:371)
> > > ...
> > > 
> > > Any advice would be much appreciated!
> > > 
> > > Best,
> > > Sandro
> 
> 



Re: trunk build does not work // was: mountByFS does not work

2016-11-15 Thread Sandro Boehme

Hi Julian,

thanks for your feedback!

Am 15.11.16 um 10:47 schrieb Julian Sedding:

Hi Sandro

org/apache/sling/jcr/base/LoginAdminWhitelist looks like your checkout
or snapshots on your machine are not quite up to date. I removed the
I'am at the latest trunk, did 'mvn clean ...' and created a new sling 
home with the '-c' option at server startup. What else do I have to do 
to be up to date?



method AbstractSlingRepositoryManager.getLoginAdminWhitelist(), so
maybe you need to first build a local snapshot of bundles/jcr/base and
then try building bundles/jcr/oak-server again.

What version of jcr/base is compatible with which version of jcr/oak-server?



Sorry for the inconvenience. I should soon be done with refactoring
these modules.

Does that mean there is nothing I can do right now?

Best,

Sandro



Regards
Julian


On Mon, Nov 14, 2016 at 9:50 PM, Sandro Boehme  wrote:

Hi Robert,


It took me a while to get the trunk to build. I had to disable the
Rat
checks and update the dependency to org.apache.sling.jcr.base
version
2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
Why isn't there a build job anymore for trunk? Should I build Sling
differently than described at the website [1]?



The context as is at [2]. Basically, the big 'trunk' jobs were too slow
and flaky. Now we have many small jobs, with none of them slow and a
couple of them flaky :-)

But the reactor build should work nonetheless.

I currently can't build it due to repository.apache.org timeouts, can
you post any errors that you get?

Thanks,

Robert


[2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Jenkins+Se
tup



The dependency to oasling.jcr.base 2.4.x has been changed by Julian in the
meantime to prepare for the oak.server-1.1.2 release.
So I could build the trunk a few minutes ago.
But the server doesn't seem to start completely. When accessing
http://localhost:8080 I get an HTTP 503 saying
"Problem accessing /. Reason:
AuthenticationSupport service missing. Cannot authenticate request."
138 bundles are available and active.

The error log shows me this error:
14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManager start:
Uncaught Throwable trying to access Repository, calling stopRepository()
java.lang.AbstractMethodError:
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
at
org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministrative(AbstractSlingRepository2.java:373)
at
org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loader.java:326)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerNamespaces(Loader.java:213)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerBundleInternal(Loader.java:225)
at
org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(Loader.java:158)
at
org.apache.sling.jcr.base.internal.loader.Loader.(Loader.java:75)
at
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(AbstractSlingRepositoryManager.java:371)
...

Any advice would be much appreciated!

Best,
Sandro






Re: trunk build does not work // was: mountByFS does not work

2016-11-15 Thread Julian Sedding
Hi Sandro

org/apache/sling/jcr/base/LoginAdminWhitelist looks like your checkout
or snapshots on your machine are not quite up to date. I removed the
method AbstractSlingRepositoryManager.getLoginAdminWhitelist(), so
maybe you need to first build a local snapshot of bundles/jcr/base and
then try building bundles/jcr/oak-server again.

Sorry for the inconvenience. I should soon be done with refactoring
these modules.

Regards
Julian


On Mon, Nov 14, 2016 at 9:50 PM, Sandro Boehme  wrote:
> Hi Robert,
>
>>> It took me a while to get the trunk to build. I had to disable the
>>> Rat
>>> checks and update the dependency to org.apache.sling.jcr.base
>>> version
>>> 2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
>>> Why isn't there a build job anymore for trunk? Should I build Sling
>>> differently than described at the website [1]?
>>
>>
>> The context as is at [2]. Basically, the big 'trunk' jobs were too slow
>> and flaky. Now we have many small jobs, with none of them slow and a
>> couple of them flaky :-)
>>
>> But the reactor build should work nonetheless.
>>
>> I currently can't build it due to repository.apache.org timeouts, can
>> you post any errors that you get?
>>
>> Thanks,
>>
>> Robert
>>
>>
>> [2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Jenkins+Se
>> tup
>
>
> The dependency to oasling.jcr.base 2.4.x has been changed by Julian in the
> meantime to prepare for the oak.server-1.1.2 release.
> So I could build the trunk a few minutes ago.
> But the server doesn't seem to start completely. When accessing
> http://localhost:8080 I get an HTTP 503 saying
> "Problem accessing /. Reason:
> AuthenticationSupport service missing. Cannot authenticate request."
> 138 bundles are available and active.
>
> The error log shows me this error:
> 14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
> org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManager start:
> Uncaught Throwable trying to access Repository, calling stopRepository()
> java.lang.AbstractMethodError:
> org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
> at
> org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministrative(AbstractSlingRepository2.java:373)
> at
> org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loader.java:326)
> at
> org.apache.sling.jcr.base.internal.loader.Loader.registerNamespaces(Loader.java:213)
> at
> org.apache.sling.jcr.base.internal.loader.Loader.registerBundleInternal(Loader.java:225)
> at
> org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(Loader.java:158)
> at
> org.apache.sling.jcr.base.internal.loader.Loader.(Loader.java:75)
> at
> org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(AbstractSlingRepositoryManager.java:371)
> ...
>
> Any advice would be much appreciated!
>
> Best,
> Sandro


java.lang.AbstractMethodError: org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdminWhitelist() - trunk build does not work // was: mountByFS does not work

2016-11-14 Thread Sandro Boehme

Am 14.11.16 um 23:02 schrieb Robert Munteanu:

On Mon, 2016-11-14 at 21:50 +0100, Sandro Boehme wrote:

Hi Robert,


It took me a while to get the trunk to build. I had to disable
the
Rat
checks and update the dependency to org.apache.sling.jcr.base
version
2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
Why isn't there a build job anymore for trunk? Should I build
Sling
differently than described at the website [1]?


The context as is at [2]. Basically, the big 'trunk' jobs were too
slow
and flaky. Now we have many small jobs, with none of them slow and
a
couple of them flaky :-)

But the reactor build should work nonetheless.

I currently can't build it due to repository.apache.org timeouts,
can
you post any errors that you get?

Thanks,

Robert


[2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Jenkin
s+Se
tup


The dependency to oasling.jcr.base 2.4.x has been changed by Julian
in
the meantime to prepare for the oak.server-1.1.2 release.
So I could build the trunk a few minutes ago.
But the server doesn't seem to start completely. When accessing
http://localhost:8080 I get an HTTP 503 saying
"Problem accessing /. Reason:
AuthenticationSupport service missing. Cannot authenticate request."
138 bundles are available and active.

The error log shows me this error:
14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl]
org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManager
start: Uncaught Throwable trying to access Repository, calling
stopRepository()
java.lang.AbstractMethodError:
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdmi
nWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;


(snip)

Seems like a mismatch between bundle versions. I think that this
class/config was moved to oak-server recently. Just to be sure, can you
navigate to launchpad/builder and:

- remove any existing 'sling' directory
- run `mvn clean verify`
- run `java -jar target/org.apache.sling.launchpad-9-SNAPSHOT.jar`

Robert



I have created new home folders with the -c option already. I followed 
your instructions anyways but I get the same error.
BTW: org.apache.sling.launchpad.SmokeIT fails with the same error 
(...AuthenticationSupport service missing...).


I agree that it might be a config error. The following bundles are 
active as seen in /system/console/bundles:

- org.apache.sling.auth.core 1.3.20
- org.apache.sling.jcr.oak.server 1.1.1.SNAPSHOT
- org.apache.sling.jcr.base 2.4.1.SNAPSHOT
I also switched to org.apache.sling.jcr.oak.server 1.1.3.SNAPSHOT by 
changing /launchpad/builder/src/main/provisioning/oak.txt. But the error 
is still there.


Best,

Sandro


Re: trunk build does not work // was: mountByFS does not work

2016-11-14 Thread Robert Munteanu
On Mon, 2016-11-14 at 21:50 +0100, Sandro Boehme wrote:
> Hi Robert,
> 
> > > It took me a while to get the trunk to build. I had to disable
> > > the
> > > Rat
> > > checks and update the dependency to org.apache.sling.jcr.base
> > > version
> > > 2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
> > > Why isn't there a build job anymore for trunk? Should I build
> > > Sling
> > > differently than described at the website [1]?
> > 
> > The context as is at [2]. Basically, the big 'trunk' jobs were too
> > slow
> > and flaky. Now we have many small jobs, with none of them slow and
> > a
> > couple of them flaky :-)
> > 
> > But the reactor build should work nonetheless.
> > 
> > I currently can't build it due to repository.apache.org timeouts,
> > can
> > you post any errors that you get?
> > 
> > Thanks,
> > 
> > Robert
> > 
> > 
> > [2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Jenkin
> > s+Se
> > tup
> 
> The dependency to oasling.jcr.base 2.4.x has been changed by Julian
> in 
> the meantime to prepare for the oak.server-1.1.2 release.
> So I could build the trunk a few minutes ago.
> But the server doesn't seem to start completely. When accessing 
> http://localhost:8080 I get an HTTP 503 saying
> "Problem accessing /. Reason:
> AuthenticationSupport service missing. Cannot authenticate request."
> 138 bundles are available and active.
> 
> The error log shows me this error:
> 14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl] 
> org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManager 
> start: Uncaught Throwable trying to access Repository, calling 
> stopRepository()
> java.lang.AbstractMethodError: 
> org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdmi
> nWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
> 
(snip)

Seems like a mismatch between bundle versions. I think that this
class/config was moved to oak-server recently. Just to be sure, can you
navigate to launchpad/builder and:

- remove any existing 'sling' directory
- run `mvn clean verify`
- run `java -jar target/org.apache.sling.launchpad-9-SNAPSHOT.jar`

Robert


trunk build does not work // was: mountByFS does not work

2016-11-14 Thread Sandro Boehme

Hi Robert,


It took me a while to get the trunk to build. I had to disable the
Rat
checks and update the dependency to org.apache.sling.jcr.base
version
2.4.3-SNAPSHOT in org.apache.sling.jcr.oak.server.
Why isn't there a build job anymore for trunk? Should I build Sling
differently than described at the website [1]?


The context as is at [2]. Basically, the big 'trunk' jobs were too slow
and flaky. Now we have many small jobs, with none of them slow and a
couple of them flaky :-)

But the reactor build should work nonetheless.

I currently can't build it due to repository.apache.org timeouts, can
you post any errors that you get?

Thanks,

Robert


[2]: https://cwiki.apache.org/confluence/display/SLING/Sling+Jenkins+Se
tup


The dependency to oasling.jcr.base 2.4.x has been changed by Julian in 
the meantime to prepare for the oak.server-1.1.2 release.

So I could build the trunk a few minutes ago.
But the server doesn't seem to start completely. When accessing 
http://localhost:8080 I get an HTTP 503 saying

"Problem accessing /. Reason:
AuthenticationSupport service missing. Cannot authenticate request."
138 bundles are available and active.

The error log shows me this error:
14.11.2016 21:30:58.119 *ERROR* [OsgiInstallerImpl] 
org.apache.sling.jcr.oak.server.internal.OakSlingRepositoryManager 
start: Uncaught Throwable trying to access Repository, calling 
stopRepository()
java.lang.AbstractMethodError: 
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.getLoginAdminWhitelist()Lorg/apache/sling/jcr/base/LoginAdminWhitelist;
	at 
org.apache.sling.jcr.base.AbstractSlingRepository2.loginAdministrative(AbstractSlingRepository2.java:373)
	at 
org.apache.sling.jcr.base.internal.loader.Loader.getSession(Loader.java:326)
	at 
org.apache.sling.jcr.base.internal.loader.Loader.registerNamespaces(Loader.java:213)
	at 
org.apache.sling.jcr.base.internal.loader.Loader.registerBundleInternal(Loader.java:225)
	at 
org.apache.sling.jcr.base.internal.loader.Loader.registerBundle(Loader.java:158)

at 
org.apache.sling.jcr.base.internal.loader.Loader.(Loader.java:75)
	at 
org.apache.sling.jcr.base.AbstractSlingRepositoryManager.start(AbstractSlingRepositoryManager.java:371)

...

Any advice would be much appreciated!

Best,
Sandro