Should I configure the lucene index during the migration from CQ5 to AEM6.1?

[michael_lemler]

The http://docs.adobe.com/docs/en/aem/6-1/deploy/platform/queries-and-indexing.html documentation tells me that "Oak does not index content by default". This it seems that it was indexed by default with Jackrabbit 2 I wonder how I must configure the indexes after the migration to AEM 6.1 (with Oak) to have quite the same search performance as with Jackrabbit 2? The https://docs.adobe.com/docs/en/aem/6-1/deploy/upgrade.html documentation doesn't say anything about in indexes.

Also I wondered that within the new blank AEM 6.1 instanze I found 59 child nodes below the oak:index node. Amongst others a configuration of the "Lucene Full Text Index". Since "Oak does not index content by default" I would have expected not even to have a oak:index node at all (by default). So are there already indexes configured within AEM 6.1?

And finaly the question how often/fast the index rebuilds/update after changes in the content. I see that I can trigger the reindex by setting the reindex property to true. But we have some automated tests which create content and afterwards search for it. Of cause I am aware that the (Lucene Full Text) index builds asynchronous and so the test tries to find the content for a certain time before failing. But is there a way to know if the index was updated? Maybe at least a note within the log files?

Thanks in advance for the help and best regards

Michael

[Jörg_Hoh]

Hi,

regarding the "Oak does not index content by default": Yes, that is true, you need to provide specific indexes to actually find your content (in case of fulltext). In JR2 there was a default lucene config shipped with JR2, which indexed everything.

AEM 6.x comes with many indexes already created, which are required for the proper (and performant) working of the AEM 6.x application. If you have specific queries in top you need to provide the necessary index definitions by yourself.

I am not aware of a log message, when an async index has been updated. Maybe you need to set the respective logger to DEBUG.

kind regards,
Jörg

[Lokesh_Shivalingaiah]

Hi Micheal,

It would come with some basic index which it needs. But you will have to look at all the queries and based on your queries and constraints you may have to create certain indexes specific to some properties or node types.

[michael_lemler]

Hi Jörg,

thanks for the answer. So if I get it right I do not need not need to add any indexes while migration if we haven't added any query specific indexes before the migration?

In particular I ask for the "Lucene Full Text Index" which seems already to be configured under /oak:index/lucene by default?

Beste regards

Michael

[Jörg_Hoh]

Hi Michael,

From a functional point of view no index is required. It just takes longer :-)

So you can add all your custom indexes after the migration has been performed and when you re-deploy your application.

Jörg

[michael_lemler]

Hi Jörg,

thanks again. But I really find it confusing that the documentations tells that there are no indexes by default. Even if I look at the Jackrabbit Oak documentation it is not consistent at all. Like on http://jackrabbit.apache.org/oak/docs/query/query.html. At the beginning is says

"Oak does not index content by default as does Jackrabbit 2."

But later on I found a passage

"By default, the following indexes are available"

Does this mean Oak does indexes by default but different than Jackrabbit 2? I find this very confusing but as result I will take that some indexes, like the Lucene full text index are avaiable by default, and own indexes are only neccessary to optimize custom queries.

Thanks and regards

Michael

[Jörg_Hoh · Accepted solution]

Hi Michael,

The statement "Oak does not index by default as does JR2" pertains only to OAK (as Apache "product"). AEM contains Oak plus indexes.

On the Oak page you linked this statement is indeed a bit strange. Can you report this on a Oak list (maybe oak-user)?

thanks,
Jörg