Apologies for absence
Alison had sent her apologies.
minutes of the last meeting.
These were accepted.
Reports from units.
George reported that Julieta believes that the air conditioning in Appleton Tower is now working. There is still an issue with the temperature sensor in the Buccleuch Place machine room. The sensor is working but it is set at too high a temperature for the position that it is in within the room. The next temperature setting below its current value is too low. It may be possible to move the sensor or introduce a fan to make the temperature within the room more uniform.
Two switch cards have failed and been replaced. One additional switch card s still to be swapped out because it has a faulty port on it.
Toby and Simon have sorted out two LDAP problems (with ldapreplicate and syncdbtoldap) that had been caused by a multi-threading patch having been overlooked. Once the patch to OpenLDAP had been applied the two faults disappeared. There is still one LDAP related issue remaining to do with the interrelationship between LDAP and updaterpms (see the Managed Platform Unit's report below).
The FC3 laptop AFS problem reported by Lindsey has been investigated by George and it turns out that the problem of not being able to log in (or in fact taking an inordinately long time to log in) does not occur when the laptop is disconnected from the network but when it is connected to the wireless network. George has diagnosed two problems and developed fixes for each which he will be testing very shortly. The same fixes should, in theory, work for FC5 as well.
Alastair has done part of the work of making updaterpms work on a 64-bit enabled platform and this has allowed the other 64-bit work in the project to continue again.
Alastair and Craig have met to discuss some details of the proposed new version of rpmsubmit that will use AFS instead of NFS. They have come up with some decisions about how the rpm repositories will map onto AFS volumes and how the apache web server will access the rpms from AFS.
Chris has been doing more work on machine classification so as to take advantage of the new folding LCFG status page.
The LCFG export server is now providing headers for the stable releases via rsync.
Achilles has been labouring under a very high rsync load; all the slave servers were syncing frequently and all at the same time. This has now been changed so that they don't sync so frequently and at different times.
Alastair reported that they are considering introducing the releases mechanism for FC3 machines in October.
Alastair has found and fixed some bugs in the fstab component that might potentially have been hit under unusual circumstances.
Recently we have hit some problems when machines have been booting that some rpms being installed by updaterpms have tried to set the ownership of files to accounts that are only defined in LDAP. Since LDAP has not started at the time that updaterpms runs the lookup of the account name will fail. However unlike the behaviour under FC3 nsswitch now hangs instead of silently failing and so the boot process itself also hangs. Tim pointed out that it is not always possible to avoid this situation by defining accounts via the auth component (which would be the normal way of addressing this problem).
Research and Teaching.
Tim reported that all the teaching packages are now available on FC5 apart from the licensed pieces of software Maple and SICStus Prolog.
Tim has been working on the web services project and Stephen Gilmore is happy with the proposed implementation plan.
They have inherited the responsibility for the emacs/xemacs environment and Graham has been doing some work on this. He is continuing to set up the postgresql server for the new teaching database server and developing a kx509 authenticated web page which students can use to request a database account..
Iain and John have got a Condor pool working under FC5. Further expansion of the pool to open lab machines is waiting on the inclusion in a stable DICE release of a kernel patch that makes USB keyboard and mouse events visible to user space processes - this should be fairly soon as Alastair has already created the kernel patch..
Rosemary has been setting up her machine as a test database server.
John has been sorting out backups for Frank Keller's Eye Tracking project.
Archie has been rebuilding the project lab and a local Active Directory service. He has also been continuing to try to get ISE Webpack from Xilinx to work on linux, so far without success (a fall back will be to use Windows machines).
Craig reported that they have implemented the new bugzilla structure and unit managers were invited to move bugs across to the new structure.
On Saturday Craig and Neil carried out the rebuilding of the ATABoy using the new chassis that had been sent previously. They also upgraded the firmware of the ATABeasts and SATABoy along with the cold spares. The Sun servers attached to the SAN rebooted without any problem but the Linux servers connected to the SAN had problems seeing the disks because the label for each of the devices had changed following the firmware upgrade and this tripped up udev. Craig and Neil sorted this out despite finding themselves in a chicken and egg situation. Neil rebooted all the other linux servers and probably as a consequence there have been no subsequent calls from users unable to access their home directories (because of stale file handles etc). The failed system disk on phoenix was replaced the day after the last operational meeting (27th July).
Having finally got permission from the administrative staff to move their home directories over to their new admin samba server this has now been done. There are still a few shared areas that also need to be moved across. Ironically the Science Support team have recently built an MDP package for the Windows AFS client which means that it would be possible to move home directories of admin staff on MDP machines across onto AFS file space and drop samba altogether.
In October the unit will be taking a new look at how to provide web services.
About 220 desktop machines have now been upgraded to FC5, about 155 of these in the student labs. The remaining 100 or so lab machines will be upgraded during the week commencing 28th August (after the deadline for the MSc project hand-in). More problematic is the remaining approximately 435 staff/PhD machines that are still to be upgraded.
Ken has done some more work on evaluating Ruby on Rails as a suitable framework in which to create the Development Meeting Software. It looks very promising. He has also needed to spend quite a bit of time developing new rules and procedures to handle resit examinations for the ITO database. He will be working with Rosemary on this once he has completed the development and testing.
Morna reported on MDP developments:
Roger reported that the local LaTeX style files used some constructs that do not work with the new version of tetex. The style files have been modified and he hopes that it should be possible to distribute local LaTeX rpms by the end of this week or next.
There was none.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh