Il giorno 02/mag/2012, alle ore 05.52, Antony Pulicken ha scritto: > ok..Got it now..Thanks..! > > Can you please add this as a feature request? I feel this is not so difficult > to be handled in the propagation manager and will be very handy for the > customers, especially when there are more than 3 target systems. Syncope > should probably do the following: > > 1. handle the request > 2. update syncope database and return an handle result at request time > (probably with a transaction id) > 3. provision to all the target systems should happen asynchronously > 4. can provide a configurable notification mechanism to notify the user/admin > by mail in case any of the target systems updates fails > 5. rest of the things will work as it is. The client should query the status > using the transaction id explicitly if to get the detailed status updates of > the transaction > > This is kind of critical for us as we are having major performance issues > when we have more than 2-3 target systems. Please let us know your thoughts.
Hi Antony, as you can see reading the roadmap we have an improvement of the propagation mechanism that should increase the performance. Btw, your suggestion seems to be good as well: we could add the possibility to execute (configurable) propagations asynchronous as you described above. Thank you for your contribution. Regards, F. > On Mon, Apr 30, 2012 at 3:38 PM, Fabio Martelli <[email protected]> > wrote: > > Il giorno 30/apr/2012, alle ore 12.04, Antony Pulicken ha scritto: > >> I'm confused now. What do you mean by 'external' resource? Is it different >> from the resource that we normally configure in Syncope? > > The actual db table, for instance. > External resources are all the physical resources linked by connectors. > >> On Mon, Apr 30, 2012 at 3:14 PM, Fabio Martelli <[email protected]> >> wrote: >> >> Il giorno 30/apr/2012, alle ore 10.47, Antony Pulicken ha scritto: >> >>> Thanks Fabio. I thought Syncope will return immediately (probably after >>> updating Syncope database), and will do rest of the propagations to the >>> configured target systems in an asynchronous manner. >>> >>> Anyway, where can I find a sample connector that has implemented the >>> TWO_PHASE protocol as you mentioned below? >> >> I'm sorry Antony, I wrote "connector" but I meant "external resource" ... >> probably this behavior could be implemented by the connector as well but, >> from my point of view, the right place is the resource (if possible). >> >> Unfortunately I've not samples for you. >> Personally I supported some developers last year to develop a soap external >> resource to be linked TWO_PHASE by Syncope but I have not held any code to >> be shared. >> >> Regards, >> F. >> >>> Regards, >>> Antony. >>> >>> Do you have any sample code >>> >>> On Mon, Apr 30, 2012 at 1:13 PM, Fabio Martelli <[email protected]> >>> wrote: >>> >>> Il giorno 30/apr/2012, alle ore 08.32, Antony Pulicken ha scritto: >>> >>>> Hi Fabio, >>>> >>>> I have configured the 'propagation mode' in all the connectors/resources >>>> as mentioned in the blog, but still the provisioning to target systems are >>>> happening synchronously. The control comes back only after the updates are >>>> propagated to all the target systems. Can you please tell what I could be >>>> missing? Do we need to use a diff REST url for this ? Please let me know. >>> >>> Hi Antony, >>> please, consider that connector must implement TWO_PHASE protocol to get a >>> like "asynchronous" behavior: it must >>> 1. handle the request >>> 2. return an handle result at request time >>> 3. process information in background >>> 4. aware syncope about the result >>> >>> At step 1 Syncope submits the request and goes ahead with other activities >>> (propagation status SUBMITTED/UNSUBMITTED). >>> At step 4 Syncope receives result notification in background and change the >>> relative propagation task status. >>> >>> If you were thinking to something more pretty asynchronous you have to >>> customize the propagation mechanism. >>> >>> Regards, >>> F. >>> >>>> Thanks and Regards, >>>> Antony. >>>> >>>> On Tue, Apr 3, 2012 at 2:58 PM, Fabio Martelli <[email protected]> >>>> wrote: >>>> >>>> Il giorno 03/apr/2012, alle ore 11.21, Antony Pulicken ha scritto: >>>> >>>>> Hi, >>>>> >>>>> Can you please let me know the difference between the Propagation modes >>>>> 'ONE_PHASE' and 'TWO_PHASES' while creating a resource? Similarly can you >>>>> explain the connector capabilities 'ONE_PHASE_CREATE', >>>>> 'TWO_PHASE_CREATE', 'ONE_PHASE_UPDATE' etc stands for? Do we have any >>>>> documentation for these ? >>>> >>>> Propagation result can be returned at request time (ONE_PHASE) or after, >>>> by calling rest interface at a certain URL (TWO_PHASE). >>>> TWO_PHASE implies: >>>> 1. request handle >>>> 2. data processing >>>> 3. asynchronous result notification >>>> >>>> https://cwiki.apache.org/confluence/display/SYNCOPE/Propagation+mode >>>> >>>> >>> >>> >> >> > >
