Apologies for absence
Craig and Toby had sent their apologies.
minutes of the last meeting.
These were accepted.
Report from Computing Executive Group
This item was not taken.
Reports from units.
George confirmed that they would be splitting the KDC master from the LDAP master (both currently on basilisk). The KDC will be moved to barrett. George has not yet finalised the specification for the hardware to run the LDAP master on.
The switch at Buccleuch Place that is showing a warning light has been investigated but there are no errors reported in any of the logs. George will check to see whether the power fluctuation 11 days ago caused the switch to reboot and whether the warning light is still lit. Thanks to the power outage at Forrest Hill 11 days ago the switches there are now running the new firmware after the switches all rebooted. All routers appear to be stable using this version of the kernel.
There is a memory leak in the SNMP daemon which has caused some machines to run out of memory. George has set up cron jobs to restart the lcfg-snmp component (which will restart the daemon).
One of the recently purchased switches has power over ethernet capability. This has been tested on the JCMB wireless access points which can now be power cycled remotely using this facility.
Iain reported that the console servers that are still at FC3 (one at Appleton Tower, two at Buccleuch Place and one at Forrest Hill) will all be upgraded to FC5 in the next 10 days. The Dell GX 110 machines will be replaced by newer HP hardware.
Chris reported that the new kernel that was currently in the development release still had some AFS related bugs and would not be used for testing or stable releases as had originally been hoped.
Stephen's work on the Inf FC6 Desktop project is going well and the project is ahead of schedule. One can now install an FC6 LCFG managed machine from a CD or via PXE and work on the DICE level headers is almost complete. There is also support now for the new Select PC hardware under FC6. Stephen will be mailing computing staff shortly about an alpha release of the LCFG managed FC6 desktop.
Research and Teaching.
The new version of ingres (ingres-2.6) has been crashing with segmentation faults intermittently. Rosemary and Tim have discovered queries that will definitely precipitate this fault. When the query came via the TEC interface the fault occurred but when it came via the sql command line interface no fault occurred. The root cause was found to the be elfingrs process which despite having been apparently ported to FC5 had not in fact been recompiled under FC5. The recompilation of the elfingrs binary appears to have solved the problem (the new elfingrs has been installed on the Informatics database server since last Friday).
Graham has been working on the HiGraph project, Iain has increased the number of nodes in the 64 node beowulf to 69 and has also started on the GridEngine scheduler project. Archie has been looking at Bluetooth support from DICE for talking to the ARM modules in the mobile robots. He can now use ssh to connect to these remote modules.
Neil reported that they had installed the new shezhu server running FC5 on new hardware (maelcum). The main problem experienced had been moving the data from the MySQL 3.2 database to the MySQL 5 database.
Gordon has set up a test bugzilla service running on FC5.
The work on the ATABeast in Buccleuch Place (together with the replacement of its disk, turning on of LUN masking and regularising the power cabling) has now been done.
There had been a SATABoy failure last Wednesday afternoon; this was fixed by the following morning.
The Unit members had been considering introducing a new ssh service with local home directories but Simon mentioned various things that would break if this approach was adopted; it obviously needs to be given more thought.
Neil has applied the security fix that he alluded to last time.
There has been no further work done on the SAN because of pressure of other work.
Since the last Operational Meeting the User Support Unit had handled 178 new RT tickets (equivalent to about 18 per working day) and resolved 69% of them. There had been a total of 209 tickets (including both new and existing tickets) resolved over the same period.
Ken has been developing his ideas for the replacement orders/inventory system and will be presenting his plan at the next Development Meeting. He mentioned that he would be proposing to use a new XML based format for recording the initial order data. He had written perl scripts to translate in both directions between the custom format designed by Alastair about 8 years ago and the proposed XML format. He had written a Document Type Definition (DTD) for the order document and a style sheet to make the display of such documents easy on the eye.
Alison is tentatively planning on introducing the latest version of RT on Wednesday 4th April. There are significant changes to the user interface. She has informed the CSOs about these differences and will be contacting other computing staff shortly. Alison confirmed that it was much faster than the current version of RT and had a more advanced and flexible query system.
Roger has continued to work on the documentation of the installroot and the LCFG machine installation process in general. He has also been involved in the administrative side of the AFS project.
There was none.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh