Background
WebAdmin allows Library staff to run Voyager batch jobs via a web interface. It is an alternative to logging in to the server and running those jobs on the Unix command line.
WebAdmin was written by Tom Briggs while he was at Shippensburg University and was subsequently "adopted," enhanced, and distributed by Endeavor. We are certainly beholden to Tom for all the effort he put into this fine program and to Endeavor for ensuring that it has continued viability as a part of Voyager.
Our library installed WebAdmin as part of the Voyager 2000.1.3 patch upgrade, but encountered some minor glitches while getting it up and running. This is not unusual for software that started its life as open-source freeware.
I anticipate that these bugs will be fixed in the next release, but I wanted to document our installation debugging efforts so that other libraries trying to install and run this version can benefit from our experience.
Documentation
For installation and configuration, we used the following documentation:
WebAdmin (Server Utilities), Voyager Accessories Manual
Instructions for WebAdmin (part of 2000.1.3 patch docs)
Installing and Configuring WebAdmin (part of 99.1 patch docs)
The earlier documentation
Installing and Configuring WebAdmin
gives information regarding setup as a virtual host, which we did
not do and which did not seem to be required based on the
2000.1.3 installation documentation.
(In spite of that particular detail, my impression was that the 2000.1.3 patch instructions augment rather than supercede the 99.1 patch docs.)
Other than not setting up a virtual host, we did it "by the book."
Daemon problems
Our initial difficulty was that batch jobs submitted via WebAdmin were not being executed, even though we knew that the correct data was being transferred to the server, and that the webadmind daemon was running.
Our problems were due to:
1) The WebAdmin CGI was putting the job info into a file in /m1/voyager/xxxdb/tmp, and the Pwebadmind daemon startup script set the TMPDIR environment variable to /m1/voyager/tmp. So the daemon was "looking for jobs in all the wrong places."
(Thanks to Bob Trotter for alerting us to this!)
2) We had an old copy of the /m1/voyager/bin/webadmind daemon. Although we faithfully followed the 2000.1.3 patch upgrade instructions, apparently the new webadmind binary did not get copied to the voyager bin directory.
(How did we discover this? When we poked around the bin directory, we noticed that every binary except webadmind had an ".old" copy. BTW, the "version" script supplied with patch releases does not examine webadmind.)
We extracted the new webadmind out of the 2000.1.3 patch tarball and manually copied it into place, then changed the TMPDIR in Pwebadmind and restarted the daemon.
Now everything seemed to work. Kinda. Some of the simple batch jobs ran OK; unfortunately, the jobs we
most wanted the staff to be able to run (bulk importing authority records and then running catjobs 11, 12, & 13) still had a few problems. For detailed explanations, see
Bug 1: bulkimport and
Bug 2: catjob 13.
