On 03/03/2015 23:03, Alan D. Cabrera wrote:
On Mar 3, 2015, at 1:34 PM, Francesco Chicchiriccò <[email protected]> wrote:
On 03/03/2015 22:10, Alan D. Cabrera wrote:
On Mar 3, 2015, at 1:03 PM, Francesco Chicchiriccò <[email protected]> wrote:
On 03/03/2015 20:41, Alan D. Cabrera wrote:
On Mar 2, 2015, at 11:51 PM, Francesco Chicchiriccò <[email protected]> wrote:
Hi Alan,
this seems very weird: could you please check which are the 4 files failing the
RAT check under common/?
I can't verify how the build is working under Mac OS X - neither Jenkins nor
Travis are reporting troubles: it seems you're building from the 1_2_X branch
or one of 1.2 tags, right?
The 1_2_X branch.
Unapproved licenses:
/Users/acabrera/asf/syncope/common/rest-api/target/apidocs/package-list
/Users/acabrera/asf/syncope/common/rest-api/target/apidocs/script.js
/Users/acabrera/asf/syncope/common/rest-api/target/javadoc-bundle-options/javadoc-options-javadoc-resources.xml
/Users/acabrera/asf/syncope/common/rest-api/target/javadoc-bundle-options/package-list
Ah, now I see what happened: you first built from master, then switched to 1_2_X without
running "mvn clean": the problem occurs because 1_2_X and master share some
modules with different type, as consequence of the huge code re-organization with had as
per SYNCOPE-620.
An idea could be to explicitly add **/target/* to RAT exclusions, I guess: WDYT?
If you’re sure that there are no generated artifacts that need to have licenses
then I guess that would work.
Or we can just add a note on the build instructions page: please remember, when switching
from / to master branch, to perform a "mvn clean" before.
I get these errors even though I perform a mV clean.
I guess that you are still on the 1_2_X branch when running "mvn clean":
the whole point is that when you built something on master and then
switch to 1_2_X (or any other branch, currently), git is not able to
remove some directories, which remain stale and cause RAT to fail.
The common/rest-api directory, in particular, is a JAR module on master
but does not exist at all in other branches, being 'common' a JAR module
elsewhere.
So, from your own situation (on 1_2_X branch with some stale
directories) you should either (more generally):
git checkout master
mvn clean
git checkout 1_2_X
or simply
rm -rf common/rest-api
You could see at that point via "git status" that nothing is actually
missing.
(These things make me sometime regret the good old SVN days...).
About
java.lang.OutOfMemoryError: unable to create new native thread
you should be setting MAVEN_OPTS as suggested by
http://syncope.apache.org/building.html#Building_Syncope
<http://syncope.apache.org/building.html#Building_Syncope>
No joy. (BTW, I had previously set my memory to 2G)
It seems this problem is intermittently happening (see [1] and [2], for
example): this is probably due to the very high (probably exaggerated) number
of concurrent threads (10000) created by ConcurrencyTest tests - I believe this
number can be lowered to - say - 1000 and still keep the test effective.
Can you please check if such change makes any difference for you? See this
patch [3] for 1_2_X.
That did the trick.
Ok, I am going to apply that patch on 1_2_X and master.
So, I’m working on using Syncope to wrap all the disparate data that needs to
be managed at ASF. Where should I start? I’m thinking LDAP, podlings.xml, and
committers.txt. Thoughts?
Wow, this is worth a new thread actually!
First of all, you shouldn't need to build Syncope from source but create
your own project as explained in [4].
Then, as in any IdM project you need to identify:
1. the resources you want to deal with (LDAP, podlings.xml, and
committers.txt as you suggest above)
1.1 does a connector exist for all identified resources?
1.2 are there authoritative resources?
2. the data flows, for each resource against Syncope:
2.1 do you want to deal with users only? or you need also to take
into account groups / roles?
2.2 will Syncope be only reading / only writing / reading & writing
from that particular resource?
3. the attribute schema to define in Syncope and the attribute mapping
for each resource (how the "givenName" LDAP attribute relate to
corresponding attribute(s) on other resource(s)?)
In particular regarding connectors:
* for LDAP we rely on the well-established LDAP connector [5] but it
would be anyway useful to be aware of the actual technology: OpenLDAP?
ApacheDS? other?
* for podlings.xml and committers.txt there might be the need to built
custom connectors from scratch, unless
** you are able to report data to CSV [6] / other XML format [7] or
** you provide some (bash?) scripts to manipulate such files and
empowers the CMD connector [8] (suggested)
Regards.
[1] https://travis-ci.org/apache/syncope/builds/52704980
[2] https://travis-ci.org/apache/syncope/builds/52896946
[3] https://paste.apache.org/0slX
[4]
https://cwiki.apache.org/confluence/display/SYNCOPE/Create+a+new+Syncope+project
[5] https://github.com/Tirasa/ConnIdLDAPBundle
[6] https://github.com/Tirasa/ConnIdCSVDirBundle
[7] http://openicf.forgerock.org/connectors/index.html#XML_File_Connector
[8] https://github.com/Tirasa/ConnIdCMDBundle
--
Francesco Chicchiriccò
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC
http://people.apache.org/~ilgrosso/
