[keycloak-dev] add-user.sh overwrites wildfly one

Sat Apr 23 15:04:18 EDT 2016

We had the same kind of problem in WildFly a few years ago.  Everyone 
was used to starting the server with run.sh.  But we needed to change 
that to differentiate between standalone.sh and domain.sh.  So we made 
run.bat just print out a "This is deprecated.  Here is what you need to 
do...." message.

It's not a perfect solution, but we could do the same thing with 
add-user.sh and tell them to use either add-user-keycloak.sh or 
add-user-eap.sh.  At least you wouldn't get any support questions.

On 4/23/2016 9:06 AM, Ilya Rum wrote:
> Hello!
>
> As a new member of keycloak QA team I recently had to set up some
> clustering with domain mode.
> I was really confused when add-user.sh did not add user to jboss but
> rather created the keycloak-add-user.json.
> The worst thing was that I couldn't find any docs on adding user to
> underlying eap at all.
> Had to read the add-user.sh itself to find out what was happening.
> Even if it remains as it is, it really should be at least mentioned in
> the docs :)
>
> Have a nice day!
> Ilya Rum.
>
> On Sat, Apr 23, 2016 at 08:48:15AM -0400, Bill Burke wrote:
>> Do you care about usability at all?  Not everything can fit into nice little
>> boxes all the time.  This is going to be extremely confusing for users.  I
>> ran into it myself as I thought the jboss add-user.sh script was overwritten
>> by our distribution script by mistake.  *OF COURSE* we should have a
>> separate add-user.sh script. Even when, hopefully, JBoss can delegate to
>> Keycloak in maybe 7.1. If we are going to leverage the JBoss platform, and
>> this means the JBoss documentation too, every management function that
>> exists in JBoss should be available in Keycloak and *WORK THE SAME WAY*.  If
>> we don't change this, we're going to get a ton of support questions that
>> say: "Why doesn't add-user.sh work?"
>>
>>
>>
>> On 4/23/2016 1:29 AM, Stian Thorgersen wrote:
>>> In the future we need to secure the underlying WildFly with rhsso. In
>>> which case our add-user will add users for both Keycloak and WildFly/EAP.
>>>
>>> IMO there's going to be confusion until the above is solved no matter what
>>> we do. We'll need to document this whichever way we do it. Options are
>>> stay with what we have or rename our script. My vote goes to keep as is
>>> and document it. Then hopefully by 7.1 we can secure the WildFly bits so
>>> the problem goes away. With the other option (rename ours) there will be a
>>> problem once WildFly bits are secured by Keycloak as now the wf add-user
>>> script should no longer be used and completely removed at which point we
>>> should then rename it back. So in the long run sticking with how it is
>>> today is ideal. It's also way to late making changes now. BTW this has
>>> been around for months.
>>>
>>> On 22 Apr 2016 22:14, "Bill Burke" <bburke at redhat.com
>>> <mailto:bburke at redhat.com>> wrote:
>>>
>>>
>>>
>>>     On 4/22/2016 3:57 PM, Marek Posolda wrote:
>>>     > That's the question...
>>>     >
>>>     > For server distribution, we also have our stuff ( keycloak
>>>     subsystem,
>>>     > datasource, infinispan etc) directly declared in
>>>     "standalone.xml". On
>>>     > the other hand, for overlay distribution, we don't want to directly
>>>     > update default "standalone.xml", so we are adding our own
>>>     > "standalone-keycloak.xml". Isn't it quite similar thing?
>>>     >
>>>
>>>     Product will not have the overlay distribution.
>>>
>>>     > We can do the same for overlay and server distribution, so never
>>>     edit
>>>     > default wildfly files ( standalone.xml , add-user.sh), but
>>>     always use
>>>     > our own versions with "-keycloak" suffix. Advantage is more
>>>     > consistent. However people will need to always start keycloak server
>>>     > with "./standalone.sh -c standalone-keycloak.xml" then. Doesn't it
>>>     > sucks from the usability perspective?
>>>     >
>>>
>>>     The overlay exists because we can't distribute EAP within community.
>>>     Keycloak should be run as a separate server, so, IMO, -keycloak.xml
>>>     files should go away and overwrite standalone.xml,
>>>     standalone-ha.xml and
>>>     domain.xml
>>>
>>>     > I honestly don't know what's the best way regarding usability. AFAIK
>>>     > this was decided on mailing lists couple of months ago, but don't
>>>     > remember the exact threads...:/
>>>     >
>>>
>>>     I'm pretty adamant about this.  There will be a huge amount of
>>>     confusion
>>>     if we don't make this separation.  Wildfly/JBoss and Keycloak are hard
>>>     enough to configure as it is.
>>>
>>>
>>>     --
>>>     Bill Burke
>>>     JBoss, a division of Red Hat
>>>     http://bill.burkecentral.com
>>>
>>>     _______________________________________________
>>>     keycloak-dev mailing list
>>>     keycloak-dev at lists.jboss.org <mailto:keycloak-dev at lists.jboss.org>
>>>     https://lists.jboss.org/mailman/listinfo/keycloak-dev
>>>
>> -- 
>> Bill Burke
>> JBoss, a division of Red Hat
>> http://bill.burkecentral.com
>>
>> _______________________________________________
>> keycloak-dev mailing list
>> keycloak-dev at lists.jboss.org
>> https://lists.jboss.org/mailman/listinfo/keycloak-dev
> _______________________________________________
> keycloak-dev mailing list
> keycloak-dev at lists.jboss.org
> https://lists.jboss.org/mailman/listinfo/keycloak-dev