These were accepted
Report from Computing Executive Group
Alastair reported that he had hoped to reveal the proposed changed to CO accommodation at the meeting but that last minutes hitches prevented him from doing so. He hoped to be in a position to mail out details of the changes by Thursday afternoon (and has subsequently done so).
Reports from units.
George asked to be told about any topics which should be added to the fixing things document
Alastair asked how hosts attached to building UPSs were notified that the UPS is about to shut down. George said that this was not quite in place yet but would eventually be available via an LCFG header.
Stephen pointed out that the output from this week's profile compilation check was incorrect due to a bug in the script. This has been fixed.
Stephen asked for suggestions for further checks.
Chris announced that after much effort, he has decided to abandon efforts to get the sleep mechanism working properly on Dell hardware. Everything seems to work reliably with the new HP desktop hardware.
There were a couple of issues with last week's lab exam. There were issues with the Java library on the day before the exam and on the day of the exam itself, it was found that the Eclipse debugger wasn't working properly, apparently due to local firewall issues. It's possible that this problem has been present in previous exams but not noticed since not many people use the debugger.
It was agreed that we really need a longer testing period for exam machines though we are to some extend dependent on the academic staff testing their material sufficiently in advance of the exam date.
Stephen proposed setting up a separate LCFG release branch for exam machines which would allow a longer testing period. It was agreed that we would try this out for the next set of mock exams.
If it seems desirable, Stephen may look at widening the random time interval during which autoreboots take place. This would be similar to the changes recently made to the cron component.
Alison commented that the exam lockdown process now appears to be much more reliable. It was agreed that the ITO needs to be reminded about the 15 minute settling in period before the start of on-line exams.
Alison reported that an appointment has been made for the Waddington building CSO post. Our new colleague starts on the 14th of September and after a training period at the Forum will spend three days a week at KB.
Topics for discussion
There was a discussion on whether there would be any major obstacles to turning off AMD on the servers once the move to AFS was complete. Tim asked about NFS mounted web space, particularly ITO related material which is currently served from stumer. This shouldn't be an issue since web servers can now access AFS space and the AFS group space file tree is designed with the aim of being able to replace the AMD mountpoints with a simple symlink once the move to AFS is complete.
George pointed out that the localhome mechanism currently relies on an AMD map and asked how we would cope with local versus network homedirs if AMD goes away. It was agreed that this could probably be handled with symlinks as well. It was pointed out that there are PAM modules available which do this sort of thing.
It was agreed that everyone would consider where they make use of NFS mounted data on servers to see if there were any other objections to the removal of AMD.
Neil mentioned an issue he had encountered with rsync and AMD. It turned out that this was an example of a known (and long standing) problem with portmap.
The next Operational meeting will be on Wednesday 9th September in IF-4.31.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh