The only 'fix' was to copy ns_reg.properties and ns_idx.properties from one node to another. I didn't modify any other files in the repository directory. In general, it seems like a bad idea to manually edit the files in rep.home.
Thomas Mueller suggested an alternate (and better) fix, but I haven't be able to try it. Check the list archives for his email. Justin On Aug 14, 2010, at 12:31 PM, Carl Hall <[email protected]> wrote: > This is the only exception I see in the log which happens repeatedly, so it > could be covering something else that can't start. > > If it means anything, the repository/index is not created when this happens. > > I set the contents of jackrabbit/repository/meta/rootUUID > to cafebabe-cafe-babe-cafe-babecafebabf > (was: cafebabe-cafe-babe-cafe-babecafebabe) but that didn't seem to help > anything. > > Is there a fix for namespace indexing that I should/could try? > > > On Sat, Aug 14, 2010 at 12:17 PM, Justin Edelson > <[email protected]>wrote: > >> Is this the only error in the log? >> >> You might be running into the namespace index problem I reported a few >> weeks ago, but that had a different error. >> >> On Aug 14, 2010, at 11:31 AM, Carl Hall <[email protected]> wrote: >> >>> The ID of node 1 is 3AKAI.NYU.NODE1. The ID of node 2 is >> 3AKAI.NYU.NODE2. >>> I'll keep digging. >>> >>> On Sat, Aug 14, 2010 at 11:08 AM, Justin Edelson < >> [email protected]>wrote: >>> >>>> Perhaps you are not using a unique cluster ID? >>>> >>>> As I understand it, you should not need to manually copy anything >> between >>>> nodes. >>>> >>>> On Aug 14, 2010, at 9:10 AM, Carl Hall <[email protected]> wrote: >>>> >>>>> I've been able to cluster JackRabbit backed by Oracle but I have a >>>> question >>>>> about something I've observed. In getting things setup, using this [1] >>>>> config, the first node (against a clean Oracle schema) always comes up >>>>> clean. The second and subsequent nodes startup with a lot of these >>>> errors: >>>>> >>>>> 14.08.2010 09:04:17.504 *ERROR* [Repository Pinger] >>>>> org.apache.sling.jcr.jackrabbit.server acquireRepository: Repository >>>> problem >>>>> starting repository from >> file:/home/chall/sling/jackrabbit/repository.xml >>>> in >>>>> /home/chall/sling/jackrabbit (javax.jcr.RepositoryException: failed to >>>>> instantiate shared item state manager) javax.jcr.RepositoryException: >>>> failed >>>>> to instantiate shared item state manager >>>>> at >>>>> >>>> >> org.apache.jackrabbit.core.RepositoryImpl$WorkspaceInfo.doInitialize(RepositoryImpl.java:2052) >>>>> at >>>>> >>>> >> org.apache.jackrabbit.core.RepositoryImpl$WorkspaceInfo.initialize(RepositoryImpl.java:1997) >>>>> at >>>>> >>>> >> org.apache.jackrabbit.core.RepositoryImpl.initStartupWorkspaces(RepositoryImpl.java:535) >>>>> at >>>> >> org.apache.jackrabbit.core.RepositoryImpl.<init>(RepositoryImpl.java:366) >>>>> at >>>>> >> org.apache.jackrabbit.core.RepositoryImpl.create(RepositoryImpl.java:673) >>>>> at >>>>> >>>> >> org.apache.sling.jcr.jackrabbit.server.impl.SlingServerRepository.acquireRepository(SlingServerRepository.java:141) >>>>> ... >>>>> Caused by: org.apache.jackrabbit.core.state.ItemStateException: failed >> to >>>>> write bundle: cafebabe-cafe-babe-cafe-babecafebabe >>>>> ... >>>>> Caused by: org.apache.jackrabbit.core.state.ItemStateException: failed >> to >>>>> write bundle: cafebabe-cafe-babe-cafe-babecafebabe >>>>> ... >>>>> Caused by: java.sql.SQLException: ORA-00001: unique constraint >>>>> (SAWG_SSO_CLUSTER.DEFAULT_BUNDLE_IDX) violated >>>>> >>>>> To get around this, I copy the repository directory from the first node >>>> to >>>>> the other nodes and they are able to startup cleanly. Is copying >>>>> repository/ an expected step or am I missing something? >>>>> >>>>> 1 http://pastebin.com/Yc7PSb1n >>>> >>
