Apologies for absence
Alastair had sent his apologies.
minutes of the last meeting.
These were accepted.
Report from Computing Executive Group
This agenda item was not taken.
Reports from units.
The organisation of the switch over to barrett as the master KDC server will be discussed as soon as staff return from annual leave. The new LDAP master server hardware has arrived. It will only be installed after the KDC situation has been dealt with.
All console servers excepting the one at Forrest Hill have now been upgraded to FC5. Replacement hardware for the one at Forrest Hill needs to be sourced before that can be upgraded.
George mentioned the announcement by Sam Wilson of a new network EdNET-FV which is being introduced to a number of recently connected student residences.
Stephen reported that there was an OpenLDAP problem with the current FC6 software release. LDAP replication was being attempted using MD5 based authentication of the transfer even though we wished to use an alternative method. This was occurring because the cyrus-sasl-md5 package was still available under FC6. Under FC5 we had got around this by removing the package but there were dependencies on it under FC6 which prevented this simple solution. A new lcfg-cyrus-sasl component has been written to carry out the sasl configuration of slapd so that it doesn't use MD5.
Introduction of support for PXE installation of a stable FC6 release has been postponed until next week. A support matrix of what is supported on what hardware will be maintained and published from when Carol returns from annual leave. It is likely that Stephen will switch his desktop to an FC6 one later this week. John, Iain and Stephen have been discussing how to plan the FC6 desktop roll out project.
Last week the control over the configuration of several Solaris servers was moved over to the new header files.
Alastair has written a VMware component called lcfg-vmwarews. It manages VMware player.
The last remaining rpm slave to be running on an FC3 machine was upgraded to FC5 when Iain upgraded it (it was also a console server).
Research and Teaching.
The School database server has crashed again with a segmentation violation fault during query optimisation; so it appears that there may be another problem with the server apart from the one that was fixed recently.
The new kernel with the USB fix is out and will be installed on the majority of lab machines when they are rebooted in the next 24 hours or so. The Unit will be increasing the Condor pool once this is done. The change restricting where jobs can be submitted from has still to be made.
Graham has started work on the postgresql component. It was originally written for postgresql 7.4 but it only half works on release 8.*. Graham will be modifying it so that it can properly configure postgresql 8.1 and 8.2.
The Unit will shortly be looking at the task of porting rpms that are needed for teaching and research to FC6. It is likely that some form of automatic rebuilding of the rpms will be attempted as part of the process. However it is normally expected that at the same time as the upgrade of the operating system the teaching and research packages are also upgraded to the latest release.
Craig reported that there had been a problem with the Appleton Tower print server yet again at the weekend. It is suspected that the problem occurs when the printcap is being rebuilt from LDAP. Since the printcap should change very rarely its rebuild has now been limited to once a day at 08:00 during the week and immediately followed by a restart of the LPRng component.
With the introduction of the new EdNET-FV network (mentioned above) the Unit will be examining all the web servers that they control to determine which access control files should be modified to also allow access from the new network.
Neil has sent out a mail message to computing staff encouraging them to move their mail over to staffmail.ed.ac.uk. After he sent this it was noticed that staffmail does not support the + notation used by sendmail. IS have been informed of the problem.
File systems on phoenix, roc and stumer that were attached via fibre channel to the Appleton Tower SATABoy were unavailable for 2 hours from 15:00 on Thursday 4th April. The SATABoy had wiped the LUN masking tables. AFS, admin samba and some NFS file systems were unavailable. Craig fixed this by restoring their permissions on the LUN masking so that the servers could access the partitions. This happened immediately after Craig had created two new AFS volumes on the SATABoy, but why it happened is still unknown. A report will be sent to Nexsan as soon as more information is available.
Since the last Operational Meeting the User Support Unit had handled 135 new RT tickets (equivalent to about 15 per working day) and resolved 54% of them. There had been a total of 128 tickets (including both new and existing tickets) resolved over the same period.
The scripts for warning users of the intention to delete their old computer account and for generating scripts to actually delete the home directories and homepages directories are now in a new version of the dice-accntmgr rpm that is in the current testing release. A modified version of the create_afsuser script was also included that fixed the problem with permissions on any home directory created under AFS.
Ken has done more work on the script to extract inventory information from (a copy of) the School database and insert new records and update existing records with data from the XML format order files. He has also introduced a new order table in the copy of the School database and added new fields to the system and hostname tables.
Roger has been doing further work on documenting the installation process. He has asked members of the Managed Platform Unit for comments on the document as it stands at present.
Alison upgraded the RT server to version 3.6.1 last Wednesday (4th April).
Ken has made a slight modification to the homedir script to correct the erroneous behaviour. The rpm containing the new version of the command is now in the stable release.
Alison and Lindsey will be looking at possible new custom fields in RT that we might use for the support queue. Ken encouraged other units to consider what customisation of their own queues they might wish to carry out.
The FC5 lab machines will be rebooted during the next 24 hours or so and messages have/will be sent out to ask staff to reboot their machines as soon as convenient to pick up the new kernel.
There was none.
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh