You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@jackrabbit.apache.org by Peter Harrison <ch...@gmail.com> on 2015/11/02 05:42:11 UTC
Re: Oak Indexes
Thanks everyone for the help.
This has all become very important for performance, and so I will have a
bit of a play with the examples you have all given me.
On Fri, Oct 30, 2015 at 2:59 PM, Clay Ferguson <wc...@gmail.com> wrote:
> Thanks Torgeir! I plan to scrutinize your code asap and see if there's any
> new info I need to incorporate into
> my own app, Seeing it done two different ways is likely to help one or more
> of us! :)
> -Clay
>
>
> On Thu, Oct 29, 2015 at 8:39 AM, Torgeir Veimo <to...@gmail.com>
> wrote:
>
> > Additionally, you can configure a lucene full text index with (in the
> > initialize method)
> >
> > NodeBuilder index = IndexUtils.getOrCreateOakIndex(builder);
> > index.child("lucene")
> > .setProperty("jcr:primaryType", "oak:QueryIndexDefinition",
> Type.NAME)
> > .setProperty("compatVersion", 2)
> > .setProperty("type", "lucene")
> > .setProperty("async", "async")
> > .setProperty("reindex", true)
> > .child("indexRules")
> > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > .setProperty(":childOrder", ImmutableSet.of("nt:base"),
> Type.STRINGS)
> > .child("nt:base")
> > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > .setProperty(":childOrder", ImmutableSet.of("properties"),
> > Type.STRINGS)
> > .child("properties")
> > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > .setProperty(":childOrder", ImmutableSet.of("allProps"),
> > Type.STRINGS)
> > .child("allProps")
> > .setProperty("jcr:primaryType", "oak:Unstructured",
> Type.NAME)
> > .setProperty("name", ".*")
> > .setProperty("isRegexp", true)
> > .setProperty("nodeScopeIndex", true);
> >
> > On 29 October 2015 at 23:35, Torgeir Veimo <to...@gmail.com>
> > wrote:
> > > Here's how we initialize non-lucene indexes in a spring environment;
> > >
> > > public class RepositoryConfiguration implements DisposableBean {
> > > [...]
> > > public Repository getRepository() throws ContentException {
> > > NodeStore nodeStore = null;
> > > segmentStore = new FileStore(new File(oakRepositoryPath), 256);
> > > nodeStore = new SegmentNodeStore(segmentStore);
> > >
> > > oakRepository = new Jcr(nodeStore)
> > > .with(new LocalInitialContent())
> > > .withAsyncIndexing()
> > > .createRepository();
> > > return oakRepository;
> > > }
> > >
> > > public class LocalInitialContent implements RepositoryInitializer,
> > > NodeTypeConstants {
> > >
> > > @SuppressWarnings("unused")
> > > private static NodeState createInitialContent() {
> > > NodeBuilder builder = EMPTY_NODE.builder();
> > > new InitialContent().initialize(builder);
> > > return ModifiedNodeState.squeeze(builder.getNodeState());
> > > }
> > >
> > > @Override
> > > public void initialize(NodeBuilder builder) {
> > > // optionally set up lucene full-text index
> > >
> > > createDefinition("jcr:name", index);
> > >
> > > createDefinition("ka:assetType", index);
> > > [.. lots of other property indexes ..]
> > > }
> > >
> > > private void createDefinition(String definition, NodeBuilder index) {
> > > IndexUtils.createIndexDefinition(index, definition, true, false,
> > > ImmutableSet.of(definition), null)
> > > .setProperty("reindex", true);
> > > }
> > > }
> > >
> > > In addition you would want a counter index, set up something like
> > >
> > > NodeBuilder entry = index.child("counter")
> > > .setProperty(JCR_PRIMARYTYPE, INDEX_DEFINITIONS_NODE_TYPE, NAME)
> > > .setProperty(TYPE_PROPERTY_NAME, "counter");
> > >
> > > But I haven't been able to get oak to make use of it if configured
> > > this way yet.
> > >
> > > On 29 October 2015 at 12:48, Clay Ferguson <wc...@gmail.com> wrote:
> > >> This file shows petty much everything I know... the sum total of my
> > >> knowledge on indexes! :
> > >>
> >
> https://github.com/Clay-Ferguson/meta64/blob/master/src/main/java/com/meta64/mobile/repo/OakRepository.java
> > >>
> > >> I was only able to cobble that code together using very scant
> > information
> > >> from various sources, and I *think* it works. Unless I just broke it
> > with
> > >> my last commit!
> > >>
> > >> Best regards,
> > >> Clay Ferguson
> > >> wclayf@gmail.com
> > >>
> > >>
> > >> On Wed, Oct 28, 2015 at 9:20 PM, Peter Harrison <cheetah100@gmail.com
> >
> > >> wrote:
> > >>
> > >>> I have moved from JackRabbit 2.X to JackRabbit Oak. I know that
> > JackRabbit
> > >>> Oak does not index as much as 2.X, and I am now getting the following
> > >>> message in my query:
> > >>>
> > >>> Traversed 4000 nodes with filter ... consider creating an index or
> > changing
> > >>> the query.
> > >>>
> > >>> I have found the IndexUtils class, and some JavaDoc, but I can't see
> > any
> > >>> more detailed documentation on how it is used, or any examples.
> > >>>
> > >>> Where should I create the Index node in the tree? Does this control
> > when a
> > >>> property is indexed? Can you index only properties, or also node
> > names? Are
> > >>> node names already indexed?
> > >>>
> > >>> Thanks,
> > >>> Peter
> > >>>
> > >
> > >
> > >
> > > --
> > > -Tor
> >
> >
> >
> > --
> > -Tor
> >
>
Re: Oak Indexes
Posted by Torgeir Veimo <to...@gmail.com>.
Here's how oak's own InitialContent configures the basic indexes;
https://github.com/apache/jackrabbit-oak/blob/trunk/oak-core/src/main/java/org/apache/jackrabbit/oak/plugins/nodetype/write/InitialContent.java
On 2 November 2015 at 14:42, Peter Harrison <ch...@gmail.com> wrote:
> Thanks everyone for the help.
>
> This has all become very important for performance, and so I will have a
> bit of a play with the examples you have all given me.
>
>
>
> On Fri, Oct 30, 2015 at 2:59 PM, Clay Ferguson <wc...@gmail.com> wrote:
>
> > Thanks Torgeir! I plan to scrutinize your code asap and see if there's
> any
> > new info I need to incorporate into
> > my own app, Seeing it done two different ways is likely to help one or
> more
> > of us! :)
> > -Clay
> >
> >
> > On Thu, Oct 29, 2015 at 8:39 AM, Torgeir Veimo <to...@gmail.com>
> > wrote:
> >
> > > Additionally, you can configure a lucene full text index with (in the
> > > initialize method)
> > >
> > > NodeBuilder index = IndexUtils.getOrCreateOakIndex(builder);
> > > index.child("lucene")
> > > .setProperty("jcr:primaryType", "oak:QueryIndexDefinition",
> > Type.NAME)
> > > .setProperty("compatVersion", 2)
> > > .setProperty("type", "lucene")
> > > .setProperty("async", "async")
> > > .setProperty("reindex", true)
> > > .child("indexRules")
> > > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > > .setProperty(":childOrder", ImmutableSet.of("nt:base"),
> > Type.STRINGS)
> > > .child("nt:base")
> > > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > > .setProperty(":childOrder", ImmutableSet.of("properties"),
> > > Type.STRINGS)
> > > .child("properties")
> > > .setProperty("jcr:primaryType", "nt:unstructured", Type.NAME)
> > > .setProperty(":childOrder", ImmutableSet.of("allProps"),
> > > Type.STRINGS)
> > > .child("allProps")
> > > .setProperty("jcr:primaryType", "oak:Unstructured",
> > Type.NAME)
> > > .setProperty("name", ".*")
> > > .setProperty("isRegexp", true)
> > > .setProperty("nodeScopeIndex", true);
> > >
> > > On 29 October 2015 at 23:35, Torgeir Veimo <to...@gmail.com>
> > > wrote:
> > > > Here's how we initialize non-lucene indexes in a spring environment;
> > > >
> > > > public class RepositoryConfiguration implements DisposableBean {
> > > > [...]
> > > > public Repository getRepository() throws ContentException {
> > > > NodeStore nodeStore = null;
> > > > segmentStore = new FileStore(new File(oakRepositoryPath), 256);
> > > > nodeStore = new SegmentNodeStore(segmentStore);
> > > >
> > > > oakRepository = new Jcr(nodeStore)
> > > > .with(new LocalInitialContent())
> > > > .withAsyncIndexing()
> > > > .createRepository();
> > > > return oakRepository;
> > > > }
> > > >
> > > > public class LocalInitialContent implements RepositoryInitializer,
> > > > NodeTypeConstants {
> > > >
> > > > @SuppressWarnings("unused")
> > > > private static NodeState createInitialContent() {
> > > > NodeBuilder builder = EMPTY_NODE.builder();
> > > > new InitialContent().initialize(builder);
> > > > return ModifiedNodeState.squeeze(builder.getNodeState());
> > > > }
> > > >
> > > > @Override
> > > > public void initialize(NodeBuilder builder) {
> > > > // optionally set up lucene full-text index
> > > >
> > > > createDefinition("jcr:name", index);
> > > >
> > > > createDefinition("ka:assetType", index);
> > > > [.. lots of other property indexes ..]
> > > > }
> > > >
> > > > private void createDefinition(String definition, NodeBuilder
> index) {
> > > > IndexUtils.createIndexDefinition(index, definition, true, false,
> > > > ImmutableSet.of(definition), null)
> > > > .setProperty("reindex", true);
> > > > }
> > > > }
> > > >
> > > > In addition you would want a counter index, set up something like
> > > >
> > > > NodeBuilder entry = index.child("counter")
> > > > .setProperty(JCR_PRIMARYTYPE, INDEX_DEFINITIONS_NODE_TYPE,
> NAME)
> > > > .setProperty(TYPE_PROPERTY_NAME, "counter");
> > > >
> > > > But I haven't been able to get oak to make use of it if configured
> > > > this way yet.
> > > >
> > > > On 29 October 2015 at 12:48, Clay Ferguson <wc...@gmail.com> wrote:
> > > >> This file shows petty much everything I know... the sum total of my
> > > >> knowledge on indexes! :
> > > >>
> > >
> >
> https://github.com/Clay-Ferguson/meta64/blob/master/src/main/java/com/meta64/mobile/repo/OakRepository.java
> > > >>
> > > >> I was only able to cobble that code together using very scant
> > > information
> > > >> from various sources, and I *think* it works. Unless I just broke it
> > > with
> > > >> my last commit!
> > > >>
> > > >> Best regards,
> > > >> Clay Ferguson
> > > >> wclayf@gmail.com
> > > >>
> > > >>
> > > >> On Wed, Oct 28, 2015 at 9:20 PM, Peter Harrison <
> cheetah100@gmail.com
> > >
> > > >> wrote:
> > > >>
> > > >>> I have moved from JackRabbit 2.X to JackRabbit Oak. I know that
> > > JackRabbit
> > > >>> Oak does not index as much as 2.X, and I am now getting the
> following
> > > >>> message in my query:
> > > >>>
> > > >>> Traversed 4000 nodes with filter ... consider creating an index or
> > > changing
> > > >>> the query.
> > > >>>
> > > >>> I have found the IndexUtils class, and some JavaDoc, but I can't
> see
> > > any
> > > >>> more detailed documentation on how it is used, or any examples.
> > > >>>
> > > >>> Where should I create the Index node in the tree? Does this control
> > > when a
> > > >>> property is indexed? Can you index only properties, or also node
> > > names? Are
> > > >>> node names already indexed?
> > > >>>
> > > >>> Thanks,
> > > >>> Peter
> > > >>>
> > > >
> > > >
> > > >
> > > > --
> > > > -Tor
> > >
> > >
> > >
> > > --
> > > -Tor
> > >
> >
>
--
-Tor