Hi Caleb,

I think I need some more information from you. Could I ask you to restart tomcat and then do one import/purge cycle and send me the three logs starting from the restart (remove passwords), thank you.

Gert


On 14/10/2011, at 17.19, Caleb Derven wrote:

Hi Gert,
 
I have a few observations from the past week.
 
I deleted/ rebuilt the solr indexes and was not able to get (401)Unauthorized during the rebuild process. I then was able to successfully purge PIDs from the solr index using the gsearch REST interface. However, going through a purge/ import cycle with objects does bring back the error. (401)Unauthorized does only appear when updating the solr indexes both SQL and risearch are updated.
 
Thanks,
 
Caleb
 
From: Gert Schmeltz Pedersen [mailto:gspe@dtic.dtu.dk] 
Sent: 10 October 2011 12:14
To: Support and info exchange list for Fedora users.
Subject: Re: [fcrepo-user] Gsearch/ solr indexing issue
 
Hi Caleb,
 
I think, I can conclude that this is a problem of your Fedora installation, not GSearch. You get (401)Unauthorized from getDatastream both at command line and Fedora admin, so GSearch correctly reports the same. So I think you have to solve it in your Fedora installation.
 
Regards,
Gert
 
 
On 08/10/2011, at 12.11, Caleb Derven wrote:


Hi Gert,
 
Thanks. Yes, the username/password is set in fedora-users.xml and Ive checked the various gsearch configuration files too.
 
The curious thing is that there are over 15,000 documents in the solr index; weve only seen the failures over the past two weeks.
 
Regards,
 
Caleb
From: Gert Schmeltz Pedersen [mailto:gspe@dtic.dtu.dk] 
Sent: 07 October 2011 17:20
To: Support and info exchange list for Fedora users.
Subject: Re: [fcrepo-user] Gsearch/ solr indexing issue
 
Is the username/password in fedora-users.xml?
 
 
On 07/10/2011, at 17.42, Caleb Derven wrote:



Hi Gert,
 
The (401)Unauthorized appears when doing getDatastream from both the command line and Fedora admin.
 
Thanks,
 
Caleb
 
From: Gert Schmeltz Pedersen [mailto:gspe@dtic.dtu.dk] 
Sent: 07 October 2011 16:17
To: Caleb Derven
Cc: Support and info exchange list for Fedora users.; Gert Schmeltz Pedersen
Subject: Re: [fcrepo-user] Gsearch/ solr indexing issue
 
Hi Caleb,
 
The Fedora API-M getDatastream replies (401)Unauthorized, indicating that the user name and password in your repository.properties are not ok. Are they the same as when you do getDatastream from command line or from Fedora admin?
 
Regards,
Gert
 
 
On 07/10/2011, at 16.19, Caleb Derven wrote:




Hi Gert,
 
Thanks very much for your response. Ive attached extracts from each of the logs.
 
I can access the fedorasgearch rest pages but am unable to do updateIndex fromPid.
 
Regards,
 
Caleb
 
From: Gert Schmeltz Pedersen [mailto:gspe@dtic.dtu.dk] 
Sent: 06 October 2011 08:51
To: Support and info exchange list for Fedora users.
Subject: Re: [fcrepo-user] Gsearch/ solr indexing issue
 
Please, let us see extracts from fedora.log, fedoragsearch.log and catalina.out from around a time, when you observe such errors. Can you access fedoragsearch/rest pages and do updateIndex fromPid actions?
 
Best
Gert
 
 
On 05/10/2011, at 15.51, Caleb Derven wrote:





Hello,
 
Weve recently encountered an issue where newly ingested objects are indexed in the risearch index and the SQL database but not in solr (via gsearch). Were running fedora version 3.4.2.
 
Following messages regarding the addition of objects and the processing of datastreams to update the SQL and resource indexes the fedora.log shows the following message:
 
ERROR 2011-10-04 06:31:48.152 [http-8080-39] (AuthFilterJAAS) Login Failure: all modules ignored
 
The fedoragsearch.log has this message:
 
ERROR 2011-10-04 12:17:48,116 (RESTImpl) dk.defxws.fedoragsearch.server.errors.FedoraObjectNotFoundException: Fedora Object ucdlib:22350 not found at gsearch_solr; nested exception is:
        (401)Unauthorized
 
Ive confirmed that no authentication details have been modified in any of the gsearch configuration files (or fedora files). There have been no issues accessing fedora via the client, web administrator or APIMs.
 
To test this behaviour weve been exporting objects, purging them, then adding the object again with the ingest object function of the admin client; they have been adding successfully and are indexed in the research and the SQL database but not solr.
 
Any help or advice would be much appreciated.
 
 
Regards,
Caleb
 
<ATT00001.c><ATT00002.c>
 
<catalina_out.txt><fedoragsearch_log.txt><fedora_log.txt>
 
 
<ATT00001.c><ATT00002.c>
 
<ATT00001.c><ATT00002.c>