news

Evolution supports a vFolder (Virtual Folder). Basically a vFolder is a query that is ran against your real folders to select messages.

It's useful. What I do now, is have my mail broken up into a number of folders. Each mailing list gets its own folder. Mail from the CS department gets a folder, etc.

Now, I can create a vFolder Inbox that can search those folders for new mail. I can construct even more complicated queries. But here's the problem: with Evolutions GUI you cannot currently construct complicated boolean expressions. Either just a group of expressions combined with AND or combined with OR, but not nested (AND (OR ...) (OR ...) (AND ...))

Evolution does support a generalized S-Expr definition of rules. It's undocumented and I've finally figured it out.

I've added a vocabulary for our NMI work. This vocab will show up when you add content to the site and will allow you to select to which NMI component the post belongs.

We now have three vocabularies available: section, project, and nmi. Content can belong to any one, one of each, or none of the terms in those vocabs. It defaults to none.

Please don't be confused by this feature, all you are doing is tagging your post with an area of work it applies to (if any). In a sense, it accounts for what you spend your time on. In the end, it gives us a powerful way to recombine our data.

I've committed some scripts for the Meta Directory project. 'bind.pl' is for binding to an authoritative source. 'search.pl' is for searching two different directories and inserting an attribute from the second directory into the first directory results. Both of these scripts are intended for use with the shell backend of OpenLDAP. I have also included my slapd.conf.metadir file for reference and a test ldif file.

jason

Random unorganized notes:

# Two ways to use a cluster, either with or without using the master node for computation. We will not be using it for computation.
# The installation guide says that, to use MPI-CH for our configuration, we need to install PBS. PBS is used to allocate jobs properly with multiple users (a single user cluster wouldn't nescessarily need it)
# /share is a large filesystem, nfs shared throughout the cluster. Since, PGI needs to be accessible from all of the nodes, I'm installing PGI into /share/opt/pgi
# I installed everything (pbs, mpi-ch), including supplementary materials (sample programs, documentation)

I've got the username and password needed to retrieve our PGI license. Right now I'm waiting for approval of a new license file (this is attached to a specific HostID of a system, and for some reason, it's different now). It should be noted, if/when we change the systems hostname, we will have to request a new license file. It supposedly takes one business day to be approved.

Next up, I think I'm going to see about installing MPI and PVM seperately. It is bundled with PGI, but I think we will greatly benefit from installing it seperately (upgradability, and greater understanding of how the

The plugins discussed in a [post on ldap-devel|http://metric.it.uab.edu/pipermail/ldap-devel/2003-December/002030.html] are specific to Sun's iPlanet LDAP server. There are two plugins:

* The [Georgetown code|http://www.georgetown.edu/giia/internet2] was written in part by Michael Gettes and seems to be an early exploration of kerb/ssl hooks.
* The [Duke code|http://www.oit.duke.edu/~rob/krbdirp/] appears to be a more mature implementation ready for production that supports separation of kerb/ssl/pam hooks.

We are interested in preventing access to port 389 for authenticated binds, to reduce the security risk of passing clear-text passwords over unencrypted channels.

The suns are up and running in rack1. Morley had a little difficult rebooting after the heat-related power loss in physics but a few fscks help clear up the complaints.

Their ip addresses have been updated and the firewall rules updated to reflect the new ip addresses. Morley is now 138.26.125.36 and michelson 138.26.125.37. Their DNS entries need to be updated, so use the ip address for access in the interim.

Their firewall is very strict and only allows ssh access from select hosts. The do not respond to pings for now. LDAP is available for all hosts.

The tape backup unit has been in a funk (flashing orange) for a while now (mid-Oct). The system just didn't want to get over a timed out command. Have spent some time on it now and then since Friday, reseting this and checking that. It finally started to work again now after a successful inventory and powercycle of the host.

I checked status of tape unit and inventory check went ok, still the backup app reported unit in paused state and showed drives not responding. Powercycled windows box and backup restarted.

It seems to be running ok now without drive errors.

I also enabled web interface on the tape unit.

We've finally got our final unpacking and arranging of furniture completed. We can now move on to other projects without having to face major lab work.

We still need to do additional filing, sorting, storing work but that can progress at it's own pace.

This sanity point has been overdue since July. To be more fair, since the new furniture arrived in Oct.

We finished the loading of the racks on Friday. All the systems in the old rack are operational. The new rack still needs to be wired for power and network. The systems will then need to be powered on.

There were several challenges during the install of the systems.

* The suns were just beasts to get into place. It's worth noting that there face plates have some missing/broken screws so care should be taken if you need to open the front of the units. Grab the left face-plate holding on to the plastic "handle" from underneath. This is the left of center. Pull straight out and the units will open. If you have questions....ask.