2010/01/28

I finally checked in the code that integrates AuCatalgoing with littleware, and setup a repository with littleware's google code site. I still need to test the etd2marc and etd2proq tools after the update, and repair some regressions, but voyager-2-vufind seems to run, and the code is in good enough shape to slap together the ebsco-export tool for Jack.

2010/01/27 more ebsco-export

Integrating AuCataoging with littleware is taking longer than I thought. I finally finished replacing AuCataloging's option object (Maybe interface), user feedback notification (Feedback interface), and properties-based Guice bootstrap ( PropertiesGuice) systems with the littlware equivalents. The code compiles now, but I know I broke some of the Guice injection modules, so I need to go back and fix that and run the regressions. All this is just a long way of saying that I won't have the ebsco-export code ready to run till next week.

2010/01/25-26

Jack assigned me a task to build MARC export files from our Voyager catalog processed for import into the EBSCO discovery service that the library is beta testing. This ebsco-export program is very similar to the voyager-to-vufind tool that I worked on last year.

I'm combining the ebsco-export development with code to integrate the AuCataloging code base with the littleware library that I also maintain.

I set aside the Solaris virtual-machine build and some pending ETD server bugs util next week. I did allow myself to get sidetracked today, and read some web pages describing SAMBA's support for DFS. I verified that DFS works on our devcat linux server; works like a charm. Samba's DFS support could help support a content-management system. I'll look into it some other time.

01/19/2010 - Solaris and ETD

It was another great day in the basement of the RBD library. First, I had more success configuring the Solaris virtual machine for testing Evergreen ILS. Solaris includes support for lightweight server virtualization and migration with containers. I'm setting up a set of sparse-root zones in our VM, and I verified today that we can add new loop-back paths to the global zone via the add fs zonecfg command. That's good to know - I can install new software into the global zone, then add the software to already running child zones.

I also managed to configure the IPF firewall in the global zone to dynamically forward global zone connections to a child zone based on the connection port via a Crossbow configured virtual network.

When you combine everything we have the ability to run a Solaris zone within a Solaris Virtual Machine, and we can checkpoint and migrate that zone to another Solaris VM with the same config. So I can test a zone in a Virtual Box VM on my laptop, or a VMware server VM at the library, get everything working, then migrate the lightweight zones to an Amazon EC2 VM running in the cloud. That's the theory.

I also spent a little time updating the embargo documentation on our ETD d-space server. Embargos are an ongoing hassle. Aaron suggested we move the embargo notice to a license-like page in the submission process where the student must click to indicate he/she understands the rules. I'll take a crack at enabling the d-space license form next week to accomplish that. I'll be in California the rest of this week.

2010/01/15

Today I continued to configure the OpenSolaris system (libsol1) to which I intend install a demonstration Evergreen ILS instance. The libsol1 server runs in a VMWare server virtual machine on an unused student workstation in the cataloging room.

Anyway, the libsol1 setup is going well so far. I finally have a Solaris zone on a ZFS file system connected to the network via a Crossbow virtual network with NAT. Buzzwords are so sexy! I'm basically following the instructions from online tutorials like http://blogs.sun.com/droux/entry/private_virtual_networks_for_solaris and https://www.sun.com/offers/docs/moving_containers.pdf. I'm working now to setup IPF firewall rules.

I'm excited about libsol1's setup, because if everything works we will be able to migrate a Solaris zone running some library software from our dev environment to a production server and also to an Amazon EC2 (elastic compute cloud) server. I'm curious to see how well an EC2 hosted service performs when accessed from Auburn across the internet. The repo.lib.auburn.edu server also runs Solaris 10, so I might be able to reuse some of the tricks I'm learning there too.

Anyway - it's all fun stuff.