incubator-syncope-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Francesco Chicchiriccò <ilgro...@apache.org>
Subject Re: csvdir connector
Date Fri, 26 Oct 2012 06:50:18 GMT
On 19/10/2012 10:06, Martin van Es wrote:
> Hi Francesco,
>
> On Thu, Oct 18, 2012 at 3:23 PM, Francesco Chicchiriccò<ilgrosso@apache.org> wrote:
>> Martin,
>> a couple of thoughts:
>>
>> 1. One way to accomplish your task (i.e. upgrade the connector bundle version in
an existing connector instance) without being forced to re-define all the components named
above would have been to export the
>> configuration as XML, make the necessary correction and then re-import the modified
content.xml.
> Its good to know that path exists, I was nearly diving into the database to see where
the connector values were stored ;)
>
>> I agree that this is not "production-ready" but it mostly fits in a dev scenario.
> If it works realiable, I don't see why that wouldn't be production ready. I don't need
fancy buttons, I asked for a way to not have to reassociate thousands of users to a new resource.
This would fit the bill I guess?

Yes, but I personally don't see this feature as completely mature yet:
consider that you cannot currently select what to export or import: it's
always the *whole* db.
This means that it's fine to make some changes "under the hood" on a dev
scenario where there are possibly few users (which are by far the
highest data volume in Syncope), but in production...

>> 2. ...
>> Instead, we could allow to change the connector instance of an existing resource
> That would be the next best alternative and get my vote immediately!
> It would also allow for quick connector debugging in dev environment.

It took some time, but I've finally been able to open an issue for this:
https://issues.apache.org/jira/browse/SYNCOPE-229

> Even better: It would allow for switching between test/prod resources without having
to reassociate the synchronisation tasks... don't know if that's smart, but it'd be nice to
have the possibility.

Hum, I don't see this as a best practice: I'd rather define two separate
synchronization tasks associated respectively with test / prod resources
so that I could better track what's happening: the trace level is
customizable, for example...

Regards.

-- 
Francesco Chicchiriccò

ASF Member, Apache Cocoon PMC and Apache Syncope PPMC Member
http://people.apache.org/~ilgrosso/


Mime
View raw message