Replication Agent's Target URI

[josephabraham]

Is configuring the replication agent to point to the publish CQ instance using a dispatcher or load balance URL a supported configuration?

When I tried the same, the asset was disappearing from the replication queue and showing status "green" against an asset. However, the content was not actually going through to the publish instance. 

I do not see any activity in the error log on  the dispatcher side which indicates that the activate request never reached my publish instance. 

Please find the log snippet from the dispatcher. 

==> dispatcher.log <==

[Sun Jan 19 22:51:02 2014] [D] [1856(46926920203136)] Found farm author for nc1d-cqapps-1001.corp.apple.com:8180

[Sun Jan 19 22:51:02 2014] [D] [1856(46926920203136)] checking [/bin/receive]

[Sun Jan 19 22:51:02 2014] [I] [1856(46926920203136)] Activation detected: action=Activate [/content/dam/aos/video/published/tr/video1.mp4]

[Sun Jan 19 22:51:02 2014] [W] [1856(46926920203136)] Unable to create file /opt/local/htdocs_storedam/.stat: Permission denied

[Sun Jan 19 22:51:02 2014] [D] [1856(46926920203136)] response.headers[Server] = "Communique/2.6.3 (build 5221)"

[Sun Jan 19 22:51:02 2014] [D] [1856(46926920203136)] cache flushed

[Sun Jan 19 22:51:02 2014] [I] [1856(46926920203136)] "POST /bin/receive?sling:authRequestLogin=1" 0 13 0ms

Any suggestions??

Thanks,

Aj

[Jörg_Hoh · Accepted solution]

When you replicate from author to publish, avoid to go through dispatcher, as it doesn't work like a real proxy in this case. It will complicate your dispatcher settings a lot (allowing POST requests to potential sensitive parts of the system). Always try to reach out directly from author to publish (a "real" proxy is supported as you can configure it in the transport settings of the agent).

Jörg

[josephabraham]

Hi 

Thanks for the reply. 

I am required to have a cluster of publish instances. Hence there will be only one replication agent fanning out of the authors (another cluster). Now, if I have a direct CQ instance as the replication receiver, then there is a risk in terms of the availability if that particular publish instance goes down. In this context, should I still refrain from using the dispatcher in the Target URI?

Regards

J

[Jörg_Hoh]

If you use the dispatcher as loadbalancer and you're happy with this feature, you might want to continue to use it. Otherwise I would use a dedicated loadbalancer.

Jörg