manifoldcf-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Paul Farrell <pfarr...@funnelback.com>
Subject Re: Manifold/Alfresco seeding and security
Date Mon, 19 Oct 2015 14:32:22 GMT
Hi Muhammed/Karl,

Firstly, thank-you so much for taking the time to reply. It is very much appreciated. 

Currently I am using the AtomPub for my CMIS repository connection. I have just read something
which may shed a little light on this. The post read that change tokens are not passed via
AtomPub connections (https://forums.alfresco.com/forum/developer-discussions/alfresco-api/cmis-change-log-token-problem-using-opencmis-03282011-1758
<https://forums.alfresco.com/forum/developer-discussions/alfresco-api/cmis-change-log-token-problem-using-opencmis-03282011-1758>).
If true, this would explain why ManifoldCF may be unable to determine a change in Alfresco.

It looks like I have two possible options left open to me (correct me if I’m wrong):

1. I look to use ‘Web Services’ instead of ‘AtomPub’ for the connection mechanism
2. I upgrade ManifoldCF so that I can use the ‘Web Scripts’ connector?  (or is this the
same as the ‘Web Services’ connection mentioned above?)

Thanks again,

Paul

Paul Farrell
Senior Search Consultant
 
109-123 Clifton Street, London EC2A 4LD
T +44 (0) 207 183 6865 | funnelback.com <http://www.funnelback.com/>

UNITED KINGDOM | AUSTRALIA | NEW ZEALAND | POLAND | UNITED STATES

Connect with us: LinkedIn <http://www.linkedin.com/company/funnelback> - Twitter <https://twitter.com/funnelback>

Funnelback UK Ltd is a limited liability company registered in England & Wales. Registered
address: Zetland House 109-123, Clifton Street, London. EC2A 4LD. Company registration number:
07004264.

> On 19 Oct 2015, at 15:12, Muhammed Olgun <mh.olgun@gmail.com> wrote:
> 
> Hi Paul,
> 
> Repositories should give information to ManifoldCF when they updated. Current CMIS connector
reindex document if the lastest version of the document has changed, not updated. 
> 
> There is a change token property in CMIS specification and it should change when document
is updated so ManifoldCF can understand that document is updated but implementing change token
property is optional.  I've checked Alfresco's CMIS web site and seen that they didn't set
the change token.
> 
> I think, there is nothing we can do at this point.
> 
> 19 Eki 2015 Pzt, 15:59 tarihinde, Karl Wright <daddywri@gmail.com <mailto:daddywri@gmail.com>>
şunu yazdı:
> Hi Paul,
> 
> This looks like a bug in the CMIS connector to me; usually the document version string
the connector constructs should be adequate to detect all changes.  Can you create a ticket?
 https://issues.apache.org/jira <https://issues.apache.org/jira> , project ManifoldCF.
 Please include what version of MCF you are using here.  FWIW, this may be in fact a bug in
the Alfresco CMIS implementation, but we'll have to have some back and forth before I can
determine that for sure.
> 
> In the meantime, have you considered using the Alfresco Webscript connector?  It's the
preferred way to do Alfresco indexing, although there have been issues reported having to
do with running it on some configurations of Alfresco.  I'm not entirely sure what the problem
is there; maybe a version dependency of some kind.
> 
> Karl
> 
> 
> On Mon, Oct 19, 2015 at 7:43 AM, Paul Farrell <pfarrell@funnelback.com <mailto:pfarrell@funnelback.com>>
wrote:
> Hi Everyone,
> 
> Hoping someone may be able to advise.
> 
> I am currently using Manifold, together with a CMIS connector, to retrieve and index
content from an Alfresco repository.
> 
> All is going well apart from, what I would call, the ‘incremental crawl’.
> 
> The main issue I am having is that the modification of a document’s security settings,
in Alfresco, is not being picked up in next Manifold crawl. As an example I have a document
‘TestDoc1’ which has user A and B as Consumers. I run a crawl in Manifold and it picks
up the documents fine.  The security is set as expected. I then remove ‘User A’ from the
security of that document and re-run the Manifold crawl. User A can still see the document
in the local search engine.
> 
> It is as if Manifold is not treating the security update as a ‘modification’ and
is therefore not refreshing it. Note that if I go into the Output Connections, edit and save
the relevant output connection and then click ‘Remove all associated documents’, the next
time I crawl, the changes are picked up. It is clear that Manifold is just not updating whatever
internal record it has for this item.
> 
> Any ideas?
> 
> Many thanks.
> 


Mime
View raw message