These were accepted.
'self managed holes' in 'what' changed to 'self managed machines'.
Report from Computing Executive Group
Reports from units.
The time allowed for vacating the AT machine room after the recent changed to the alarm set-up is still 30 seconds.
It was agreed that in the case of the AT switches, the Inf Unit would choose a date after liaising with the User Support Unit to avoid deadlines, exams and the like. For KB, it was agreed that switch reboots should take place at the end of the working day to avoid interfering with backups.
Stephen noted that Richard has a virtual student lab set up useful for testing this sort of thing.
We have been experiencing problems with both the ssh servers which have been crashing at the same time early in the morning. Possible causes include a flawed cron job and malicious behaviour. Stephen is investigating.
Topics for discussion
There was a lengthy discussion which raised several useful points. It was agreed that reducing the retention period for full backups was reasonable though it was felt that a retention period of 13 months rather than 1 year would be preferable. Craig will take these proposals to CSG.
It was agreed that archives should be split off from backups and that a new development project should be created to research how this problem is handled by other organisations within and outwith the University and look at options for how the School could manage archives.
It was suggested that it would be useful to have a way of easily determining the backup/archive retention policy for each volume. At first it was suggested that a file with this information could be created at the top level in each volume but on reflection it was agreed that it would be better to have a script which looked up this information in a central database.
It was agreed that given the current level of restore requests, there was no need to introduce a better mechanism for handling restore requests at the moment.
Although the /Yesterday mechanism is useful, some thought should be given to how snapshots of home directories at different times (eg /Thismorning, /Lastweek) could be provided.
This was deferred until the next meeting
After some discussion, it was agreed that at the moment, the sole reason for profiles for self-managed machines with dynamic IP addresses was to work round deficiencies in the inventory system. Having large numbers of such profiles clog up the LCFG servers. Until the inventory system can be fixed, it was agreed that Stephen would look into creating a suitable header for self manged machines with dynamic IP addresses which would contain the required information without adversely affecting the LCFG servers.
It was agreed that this would be fairly easy to do for sys-announce but would be considerably harder for seminars which contains a significant number of people who are not directly associated with the School. It is possible that Prometheus might provide a solution for this but in any case the intention is to move Informatics mailing lists to the new Central service in the near future. It was agreed that RAT would look in to the automatic generation of sys-announce and that Craig would create a new project to handle the migration of mailing lists to the central service.
Student helpers have quite legitimately been given the "staff" role. However, they are undergraduates and having this role might give them permissions which they shouldn't really have. What is needed is a number of staff roles. It was agreed that the whole issue of roles needs sorting out and that Tim and Alison would come up with a solution for the student helper problem whilst Alison would create a development project for improving the roles system.
The next meeting will be on Wednesday 8th February in room 4.31 chaired by TBC.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh