Apologies for absence
Alison, Morna and Roger had sent their apologies.
minutes of the last meeting.
These were accepted.
Report from Computing Executive Group.
Alastair reminded the meeting that the minutes of CEG meetings are published on the web.
Julieta has resigned and will be moving to a post in MIS in mid-November. We have had permission to reappoint to this post and Ken will be managing the reappointment process.
The process to reappoint to Sarah and Charlie's posts has been progressing (see the details in the User Support Unit report below).
There will be a brain-storming meeting in late October, attended by Alastair, Craig, George and Ken, on planning for the move to the Informatics Forum. There is still some variation in procedures from site to site which will need to be replaced by a uniform set of procedures for the Forum and Appleton Tower. We should be deciding how things should be once the major moves have been completed (around Christmas 2007 perhaps) and then work back from there to plan the transition. After the initial the brain-storming session it will be important to open up the discussions to elicit input from the wider circle of computing staff (possibly via a wiki web).
Craig, George, Ken and Tim will be attending a management training course (one day per month) between October and February. Alison, Craig, George and Tim are also on the waiting list to attend training in appraisals. We will then restart the staff appraisal process for all computing staff. This should allow us to provide advice to staff on more focused development study/training. This will possibly be integrated with a relaunched series of technical talks by staff to other computing staff.
It has been agreed in principle that there should be a computing staff travel budget for attending conferences and visiting other computing sites of interest. The new budget should have some funds in the next financial year.
The Kenway report (the report of the panel for the review of computing and IT-based services 2006) will be published by the end of this week. Available members of the Computing Strategy Group will meet on Monday to discuss our response, which needs to be ready by the end of next week, so that it can be taken to a special meeting of CCITC (College Computing and IT Committee).
There was some discussion of why some parts of the CEG minutes were marked as Restricted. It was agreed that if a restricted agenda item later became unrestricted then it would be marked as such.
Reports from units.
The infrastructure secondary router at Forrest Hill, haitink, has had its faulty disk swapped out. Apart from this Julieta reported via George that everything else on the networking front was fine.
The ldap schema changes for handling AFS home directories and the monitoring system under development have now effectively been rolled out.
A new LDAP problem (capabilities not propagating correctly) has been identified and a fix developed and implemented. The fix does have a known issue but this would only be relevant if the processes were to be run on separate hosts rather than a single host, as is now the case.
It was noted that the routers in the Appleton Tower machine room did not reboot again after power was restored to them last Saturday evening. It was a case of a BIOS setting being wrong. However as it would require another reboot to correct this no action has been taken to modify the BIOS setting at this stage.
It has been decided that there are not sufficient benefits to introducing software release management for the last few months of FC3 support and so this will not happen.
Alastair has been doing some work on the updaterpms software (primarily to add support for 64-bit architecture software packages) and has taken the opportunity to add the capability to specify that rpms are to be included in the list of rpms for a machine on the basis of lcfg contexts.
There are a number of open bugzilla bugs for porting some aspects of the managed platform infrastructure to FC5 that are needed for servers running FC5. Alastair asked that anybody that finds that upgrading their server to FC5 is being blocked by one of these bugs to let the Managed Platform Unit know so that they can prioritise these actions accordingly.
Alastair gave advanced warning that the Solaris machine headers will be moved to the svn repository and that this would require coordinating with the Services Unit.
The web pages that record hardware support and header support for FC5 are currently maintained manually. It is planned that XML markup is added to the header files so that these web pages can be created automatically each night.
There was some discussion about the original reasons for using KX509 authentication on the LCFG web server.
Research and Teaching.
Graham has finished writing the documentation for users and support (management procedures) for the postgresql server and has announced these. Courses will start to use the new server from today. He has started transferring the databases off the old (RedHat 9) postgres server onto the new database server pgteach. The old postgres server will be switched off a week on Friday. Front Line Support will take over some of the user management processes once this has been signed off by the User Support Unit.
Archie and Iain have tried unsuccessfully to get the Logitech webcam purchased by Bob Fisher to work with xawtv (there appears to be a known issue with this combination). Iain has identified gmplayer as software that does work with the Logitech equipment. John Butler will be lending seven Philips webcams to Bob for the time being since these are known to work with xawtv.
Iain has had a few problems with people running large jobs on the GridEngine head node with the effect that it has denied others from using GridEngine at all. He will escalate this if the users ignore his requests to desist from doing this.
There have been various issues with the ITO database in the last few weeks mainly to do with the rollover of the session. Tim entered a few resit marks for MSc students (less work than fixing the session-lock rules and procedures). The cvs commit of the ITO web pages generated from the database have hit cvs limits on the total number of files that can be committed in a single action again and need some attention.
Tim reported continuing problems with a few pieces of teaching software under FC5 (Isabelle, ProofGeneral and SMLNJ). There had been an unannounced new release of ProofGeneral and an incompatibility between this and Isabelle. The most recent version of SMLNJ is not compatible with cwb (Concurrency Workbench) and so it has been necessary to downgrade SMLNJ to an older version. Various Eclipse UML tools had been requested for use by different courses at the last moment and have now been made available.
John has made a couple of changes to the default environment to ease the use of HUGS and one other piece of software.
John has hit problems while setting up new video conferencing equipment for AIAI. The equipment provided was faulty and had to be sent back.
John has completed the work for the eye tracker lab.
The backup server, ouroboros, has crashed a few times recently (but this has not affected the successful backing up of servers). It is thought that the crashes are caused by a faulty SCSI controller. We have a spare controller and will swap the suspect one out.
There has been a problem with the Appleton Tower print server recently. It is fixed by rebooting the server and this is all that it has been possible to do so far because the ITO had requested no prolonged breaks in the printing service at the time of preparation of reports for Boards of Examiners. It will be investigated as soon as this is possible.
Last Saturday the homepages web server ran out of memory and was locked up until the Monday.
There has been a disk failure on the ATABeast at JCMB; Craig and Neil will swap it out for a new one after this meeting.
Nexsan have asked that we swap out the chassis and controller on the ATABoy at KB. This will need to be scheduled for a weekend.
The 8 slot LT02 super-loader tape loader had been causing problems again (see earlier reports from the Services Group meeting minutes). A replacement chassis has now arrived for this equipment and it will be swapped out tomorrow.
The meeting expressed its thanks to Neil for the work that he did a week ago on Sunday to power up and boot servers in the Appleton Tower machine room following the day long power outage in the building. Only one machine, belly.inf, failed to reboot, and that wasn't providing any active services anyway.
There have been one or two occasions when our mail server has been listed in error by the real time black hole service SpamCop. This has been caused by vacation messages being automatically sent to one or other of SPamCop's spam traps following the receipt of a piece of spam that maliciously has the spam trap address as the return address.
Neil has updated the dice/options/apache1.3.h header file so that it can be used by an FC5 host.
The opportunity was taken just over a week ago when the file server roc was powered off to install a fibre channel card and interface it to the fibre channel storage in Appleton Tower.
There is now just one member of computing staff still using an NFS home directory, George (but he has a reason for still needing an NFS home directory).
Ken reported the figures (as supplied by Alison) for desktop machines that are now running FC5:
It looked as if we were still on track to upgrade the vast majority of staff/PhD machines by the end of October.
Alison had produced a long list of candidates with some help from Lindsey (and some minor input from Ken) for the two Computing Support Officer posts that had been advertised. The closing date had been this Monday. Yesterday Alison Lindsey and Ken met and carefully chose eight applicants for the short list to be interviewed on Thursday 12th October.
Roger has been liaising with the Computer Systems Biology group who are moving from the Appleton Tower level 6 to the Darwin Building temporarily. He will be acting as their first point of contact if they have any computing issues. He has also been preparing and delivering his introductory talks to students on Unix and the DICE computing environment.
Ken has been doing further work on the Development Meeting software.
There was some discussion about the different regimes that applied to the various machine rooms. It was clear that there was a need to harmonise procedures across the different sites. The Infrastructure Unit would need to look into this.
Support for Institute/research group websites
There was some discussion of the circumstances under which it was appropriate to give help with creating web pages on Institute/research group websites.
It was appropriate to give advice about publishing tools and technology.
Alastair reminded the meeting that it was not computing staff's responsibility to produce web page content on behalf of others (just as it is not our responsibility to debug other people's java code say).
If an Institute/research group was willing to pay for computing staff to produce the content of web pages then the request should be directed to the Research and Teaching Unit.
It was agreed that we should have a general multi-user machine that was running FC3 but that it would have a very limited life after services were all transferred to FC5. It was agreed that bu.inf could be used as the machine for this temporary service, and that it should have the alias fc3.login.inf. The User Support Unit would be responsible for it.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh