Expand my Community achievements bar.

SOLVED

Deploy to publisher causes felix config to stop working

Avatar

Level 4

I have a custom saml implementation with configurations in our code base for multiple publishers. Each publisher has its own configuration.

When we deploy the saml authentication fails. If I go to the custom saml felix config and simply click the Save button, authentication starts working normally.

Any ideas why this would be happening?

Thanks

Tom

1 Accepted Solution

Avatar

Correct answer by
Employee Advisor

Check the configuration of the "Apache Sling JCR Installer" in the Webconsole. Most important are in your case the "search paths" property (there should be /apps and /libs), the installation folder name regexp (".*/(install|config)$") and the Max hierarchy depth (default 4). If it's a systematic problem on all of your publishs you might override some settings there.

If that sounds ok, you need to debug the issue more systematically

* set the facility "org.apache.sling.installer.provider.jcr.impl" to DEBUG and observe the logs (use a dedicated logfile). There should be a ton of activity constantly going on.

* then add a setting and check if these DEBUG logs show any activity (they definitly should)

Jörg

View solution in original post

8 Replies

Avatar

Employee Advisor

Do you have any logs indicating a problem? If the deployment does not take effect, there should be any statements in the logs.

Jörg

Avatar

Community Advisor

Jörg Hoh

  We also have a similar kind of situation with our 6.2 servers. Any config deployed needs a restart of the server to be picked up . Some configurations , we have to go to system console and click save. We raised a daycare ticket and doesn't seem like they are able to figure out a root cause for this issue.

Thanks

Veena

Avatar

Employee Advisor

Can you check if you have any nodes below /system/sling/installer/jcr/pauseInstallation ?

Jörg

Avatar

Level 4

There are /system/sling/installer/jcr/pauseInstallation on our publishers.

Avatar

Level 4

We are on 6.2 also Veena. Sorry to hear you are having this problem also.

Avatar

Employee Advisor

If there are nodes within this folder, remove them. Then changes during runtime are picked up again.

Jörg

Avatar

Community Advisor

While doing debug session with Day Care team, they suggested the same and we tried but there was nothing in that folder Jörg Hoh

Avatar

Correct answer by
Employee Advisor

Check the configuration of the "Apache Sling JCR Installer" in the Webconsole. Most important are in your case the "search paths" property (there should be /apps and /libs), the installation folder name regexp (".*/(install|config)$") and the Max hierarchy depth (default 4). If it's a systematic problem on all of your publishs you might override some settings there.

If that sounds ok, you need to debug the issue more systematically

* set the facility "org.apache.sling.installer.provider.jcr.impl" to DEBUG and observe the logs (use a dedicated logfile). There should be a ton of activity constantly going on.

* then add a setting and check if these DEBUG logs show any activity (they definitly should)

Jörg