« April 2009 | Main | June 2009 »

May 24, 2009

Memory Leak in DSPACE

There is a memory leak in DSAPCE.
I have tried:
1) Making a more strict robots.txt file:

User-agent: *
Disallow: /browse-subject
Disallow: /browse-author
Disallow: /browse-title
Disallow: /browse-date
Disallow: /suggest
Disallow: /*/browse-subject
Disallow: /*/browse-author
Disallow: /*/browse-title
Disallow: /*/browse-date
Disallow: /image
Disallow: /feed
Disallow: /password-login
Disallow: /advanced-search

This change (especially the Disallow: /feed ) seems to have reduced cpu load, but the leak is still there.
2) A user group suggested shutting off the string cache. I tried this but it did not seem to help. 3) Also a site suggests that there is memory leak in tomcat 5.x. The site suggests that I upgrade to tomcat 6.x I haven't done this yet. 4) I will need to look in detail at output from hmap to determine where the trouble is. It is likely within the dspace app.

Control Groups:

Control Groups:

a. Managed Content (M): Datastream content is stored and managed within the Fedora repository’s persistent storage. The content can be any MIME type including XML. b. Inline XML (X): A special case of M, restricted to well-formed XML. In this case the datastream content is stored as part of the XML structure of the digital object itself and is thus included when the digital object is exported (e.g., for archival purposes). c. Externally Referenced (E): Datastream content is external to the Fedora repository and is referenced by a URL that is recorded within the digital object. The content can be any MIME type including XML. d. Redirected Content (R): Like E, but datastream content is delivered to the client without any mediation by Fedora; i.e., via an HTTP redirect. You should use this datastream type when the external content is a web page with relative links or it is streaming audio or video. The content can be any MIME type including XML.


State
"A" "I" "D" (Active, Inactive Deleted)
Fedora object type(s)
O=regular data objects, D=behavior definitions, M=behavior mechanisms
.

Install AgEcon on strip1

Install script for AgEcon on strip1

I wrote a shell script called deployAgEcon.sh. This has all the steps required to install a new version of AgEcon on strip1.

dspace.cfg config file

All information related to installing dspace on a given box is stored in the config file dspace.cfg. Below are versions for various boxes for both UDC and AgEcon: UDC on strip1
AgEcon on strip1
UDC on odin (silvi003 account)
AgEcon on odin (silvi003 account)

May 13, 2009

Eric Moore's Comments on the UDC Indexer

Eric Moore wrote a great explanation of the fields that are indexed in UDC.

May 8, 2009

New robots.txt file for dspace

I am going to use the robots.txt file below it is based on information from the dspace wiki User-agent: * Disallow: /browse-subject Disallow: /browse-author Disallow: /browse-title Disallow: /suggest Disallow: /*/browse-subject Disallow: /*/browse-author The "/suggest" corresponds to a page that sends an email to a friend.

Basic solr works

The two requests below will get a record from solr that has "hal" in the title. http://128.101.29.84:8080/solr/select/?q=hal&start=0&rows=10&version=2.2&indent=on
http://128.101.29.84:8080/solr/select/?q=hal

Using this solr schema and ustore code