Apologies for absence
There were none.
minutes of the last meeting.
These were accepted.
Report from Computing Executive Group
As Alastair was not at the meeting, this item was not taken. Gordon asked why Alastair's presence at the meeting was a requirement for this item when there were several other members of CEG present. It was suggested by Ken and Craig that the item was perhaps misnamed and that "Report from the Head of Computing" would be more appropriate. Nevertheless, the general view of the meeting was that it would be useful for a CEG member to give a brief overview of CEG's activities since the last operational meeting, particularly in view of the difficulty sometimes experienced in spotting new additions to the CEG minutes. It was agreed that Ken, as Deputy Head of Computing would be responsible for giving this report in Alastair's absence with the meeting convener taking on the task should both Alastair and Ken be absent.
Reports from units.
The general view of the meeting was that a detailed verbal report by a representative of each unit was not necessarily the most productive use of time during the meeting. It was agreed that in future each unit will submit a short written report on their activities by the end of the Monday before the meeting, linked from the meeting agenda. A representative of each unit will give a brief summary of the report at the meeting and invite questions.
Simon raised the issue of how the individual units could publicise their future strategies. It was agreed that although there was a need for a forum for such discussion, that forum was not the Operational meeting.
Research and Teaching.
Graham had met with the CISA Implementation Task Force who had expressed concerns that the forthcoming server upgrades would result in problems such as CGI scripts breaking. They had requested a timetable for when servers would be upgraded. It was agreed that although we were not yet in a position to give firm dates for when upgrades would take place, Graham would go back to the ITF with a commitment to giving a fixed amount of warning before a server upgrade takes place.
It is now possible to order GPFS for SL5 from the IBM website. That is, unless you are using the software under an academic license (as we are) in which case the option is not available. Iain will continue to monitor the situation.
George reported back on a meeting held last week concerning the proposed power shutdown in the Appleton Tower basement sometime in October. There are two possible dates for this, the 19th to the 21st or the 26th to the 28th. As yet no final decision has been made though it is believed that the later dates are the favoured ones. The intention is that a designated person will make sure that all the kit in our machine room has been shut down by 6pm on the Friday and will then open the machine room circuit breakers to make sure there is no leakage from the UPSs back into the main circuit. At some point on the Sunday, we will be notified that the work has been completed and that power has been restored. We need to decide whether we will restore services on the Sunday or will wait until the Monday morning. During the shutdown, toilets and probably lifts will be out of action although the power will still be on in the upper floors of the tower. The intention is that the EDLAN core will remain up during the work so Buccleuch Place should be unaffected by the work (other than by the loss of services located in the AT machine room of course). George pointed out that the Student Mail Service servers are locate at AT as are the central home directory servers and that consequentially these services would be unavailable to our users. The Staffmail servers are located at KB and should be unaffected. George will circulate the final program for the shutdown when it comes out. He has also created a Wiki page with information about the shutdown. There is also an older Wiki page from a previous proposed shutdown which may prove useful.
George gave an overview of the building work which is about to start in the basement of the tower, part of which is the construction of new workshops for our technicians. Our machine room should not be affected by this phase of the work. George pointed out that our fire safe in the AT basement will need to be moved. He will feed back to those responsible for the building works that the fire safe is still in use and that we require access to it. It is possible that it will be moved to another central location, not owned by the School. George has a copy of the plans for the AT basement should anyone wish to inspect them.
Toby confirmed that COSIGN will be ready to go into service once this week's stable release has propagated. Simon mentioned that access control via LDAP should be possible.
Stephen reported that all the auto-built packages have now been resubmitted to the appropriate buckets. Longer warning will be given for the removal of auto-built packages from the SL5 package list. One of the possibilities the MPU will be examining as part of the buildtools project is the provision of a build farm to simplify the submission of packages.
New FC5 and FC6 OpenAFS kernel modules will be distributed this week to deal with a number of outstanding issues. Clients will need to be rebooted using the auto-reboot mechanism developed by Chris but now the responsibility of the User Support Unit. Ken asked whether this mechanism had been tested under FC6. The consensus was that it had not and that this should be investigated as quickly as possible.
The LCFG slave servers are now making use of the apacheconf component. This means that nearly all MPU managed servers are now being monitored via Nagios (nagios.inf.ed.ac.uk). Bill asked if it would be possible to have a web page providing links to the various monitoring mechanisms (Nagios, the machine room temperature sensors etc.). It seems likely this would be a simple thing to do. Simon mentioned that he is hoped to give a talk on the School's Nagios setup to an audience from outside the School.
A large amount of LCFG documentation has been moved from various locations to wiki.lcfg.org. More will be shifted across as time permits. If anyone would like an account to edit this wiki, they should contact Stephen or Chris.
Craig reported that the test multi-function printer/photocopier/scanner is being delivered on Friday the 19th of October. It will initially be locate in Buccleuch Place. Simon will try to locate some Postscript 'torture scripts' he has to aid assessing these machines.
A disk died in the KB ATAboy last week. It was replaced by the cold spare and Nexsan have supplied a replacement cold spare.
There have been a couple of problems with web services, one with www.inf and one with homepages/wiki.inf. The www.inf problem was caused by a very old version of Internet Explorer repeatedly requesting small amounts of a PDF file causing the limit on the number of HTTPD processes which can be run on the server to be exceeded. It was suggested that Acroread rather than the browser might be the culprit in this case. The problems with homepages/wiki.inf were cause by the scripts of a user who had exceeded his webspace quote. The user in question has amended their scripts to ensure that this will not happen again. Neil is looking into installing an apache module which automatically blacklists machines making an excessive number of connections to the server.
Please see attached written report. Ken brought up the issue of visitor email accounts and presented some possible options. Not all visitors are entitled to a staffmail account and the vast majority of our visitors fall into this category. In view of our wish to move commodity mail services to staffmail, a solution will have to be found.
Future form of Operational meetings.
There was some discussion regarding what changes should be made to the Operational meeting in order to improve communication between computing staff and encourage attendance at the meeting. It was agreed that in general the meeting was more useful when a larger number of people attended. The possibility of only holding one meeting a month was raised but it was felt that this would result in too large a gap between meetings. To help stimulate discussion at the meeting, Simon suggested that a mechanism similar to that used at LISA should be used where topics for discussion are suggested and then voted upon with the most popular topics being discussed as time permits. It was agreed that a Wiki page would be set up where all computing staff can suggest topics for discussion and indicate their interest in the topics suggested. There will be a firm limit of two hours set for the length of the operational meeting, topics not discussed being carried forward to the next meeting. It was agreed that this mechanism would be put in place for the next operational meeting.
There was some discussion on how these discussions should be minuted. It was agreed that the minutes would only contain a note of the topics discussed and any decisions reached but that a mechanism would be provided allowing any member of the computing staff to add notes enlarging on any aspect of the discussion.
In order to integrate CSOs further into the meeting structure, it was agreed that there would be a new agenda item for a report on CSO issues. Alison or a designated deputy will give this report.
The next meeting will be on October 24th in JCMB 2511 and chaired by Craig Strachan.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh