These were accepted.
All seems well so 64 bit SL6 will be the default in the labs
Done, and linked from systems page
In the end Alastair took this to the teaching committee. As it turns out, it seems that the cost of adapting teaching materials to use something other than Matlab would outweigh the saving made by using another software package. We will continue to monitor the situation.
Blog articles suggested:
Report from Computing Executive Group
Link to minutes to follow.
Reports from units.
It seems that the false alarms are being triggered by the PIR sensor in the room. One theory is that the sensor is being triggered by the increased heat generated when the Beowulf revs up. Ian and Iain will investigate.
Neil asked if the move of the main nagios server to different hardware meant that the web interface to jabber could once more be made available. Toby will investigate.
In response to a question from Neil, Stephen stated that the current crop of reboot requests do not have security implications and can be delayed if desired.
Stephen announced that SL5.6 has now been released. This may be the last SL5 upgrade we do.
In response to a question from Alastair, Roger reported that the response from first semester lecturers to his request for details of packages required for teaching purposes had been good with only a couple of responses still outstanding. Roger hopes that the semester 1 packages in 64 bit form will be available in a week or so.
Craig reported that although the firmware updates requested by DotHill are categorised as urgent, the bug they fix is not actually present in the version of the firmware currently on the School's arrays.
Topics for discussion
Several points were raised. It was agreed that if a user chooses to turn off the sleep option, this should persist across reboots. Chris will check whether this is the case. Stephen observed that sleep should be automatically disabled if the header which opens holes in the firewall for the ssh daemon is included in a machine's profile.
Neil enquired about the possibility of waking a machine from sleep remotely. Investigating wake-on-LAN is on the MPU's list of things to do. IS have wake-on-LAN working and we should find out what their experience has been.
It was agreed by the meeting that we should propose the deployment of sleep on SL6 as part of the Power Saving Action Plan.
Stephen pointed out that many of the pages on the dice web site are protected by .htaccess files using IP address based restrictions. This can make it difficult to access these pages when working outwith the School. Stephen asked whether we could move to controlling access to these areas via cosign. This was generally viewed as a good idea by the meeting until George pointed out that many of these pages are important for disaster recovery and that if the cosign service was unavailable, these pages would be inaccessible. In the end, a compromise was reached whereby the existing .htaccess files would remain unchanged but apache would be configured to use a different access file (probably .sslaccess) when these areas are accessed via https and that these files would control access via Cosign. Access from within the School will therefore be unchanged (and not Cosign dependent) whereas access from outwith the School will be Cosign controlled.
Two actions emerged from this discussion, Neil will produce documentation on how to Cosign protect pages and Units will trawl through the old groups documentation tree to establish what, if any, of this documentation remains relevant.
Several helpful suggestions for changes and additions to the document were made. These will be incorporated in a new version which will be recirculated for comment.
One issue that arose as a result of this discussion was that of password security and how we can best avoid having our users use weak passwords. Mention was made of a PAM module which captures passwords entered by users and flags up passwords considered weak. It was agreed that the Inf Unit should add investigation of this module to devproj 168
Chris asked whether it would be possible to make the Turing room the default venue for all CO meetings. The general consensus of the meeting was that this was extremely unlikely.
The next meeting will be on Wednesday 13th July in room 4.31 chaired by persons as yet unknown.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh