Operational Meeting

Apologies for absence

Morna had sent her apologies.

minutes of the last meeting.

These were accepted.

Reports from units.

• Infrastructure.

  Simon has come up with a set of options so that the nss_ldap module does the right thing under most conceivable conditions. The settings have been undergoing tests and look very promising. This addresses the problem mentioned at the last meeting under the Managed Platform report

  Toby is preparing changes to the open ldap component and is investigating an undesirable mode of interaction with slapd. He hopes to be able to roll out the changes next week. Schema changes for the component will follow but are dependent on the first changes being made.

  The Infrastructure Unit is looking at ways of isolating the Kerberos KDC from the LCFG profile for the server, possibly so that any changes have to be ratified by a trusted human before they are applied.

  The sensor in the machine room at Buccleuch Place will be moved by the technicians and a fan is on order to make the temperature within the room more uniform.

  George has rolled out the fixes for the problems affecting laptop users with AFS home directories described at the last meeting. The fixes have thrown up a new problem which the AFS project team are now looking at to do with AFS and local home directories coexisting.

• Managed Platform.

  Alastair has continued the work of making updaterpms work on a 64-bit enabled platform. In future, any rpm mentioned in a package list will need to specify the architecture if it is not the same as the architecture of the machine on which the rpm is to be installed (even if the rpm has architecture noarch). There are some LCFG tools that will need to be updated to support this change.

  A problem with the gnome screen saver for people with AFS home directories has been fixed; a change has been made to gdm so that it writes a file to /tmp instead of in the home directory. A problem with the KDE screen saver, missing PAM (Pluggable Authentication Modules) configuration, has been fixed.

  Stephen has also been looking into standardising the kernel that is run on clients running a particular operating system; the default kernel will be specified in dice/options/kernel.h.

  More work has been done on the LCFG export server.

• Research and Teaching.

  Tim reported a bug in gurgle-caingres (or more specifically the ingres library that it uses) which has been thrown up by the new version of glibc when running one of the database reports. The error is:

  *** glibc detected *** free(): invalid next size (fast): 0x08d1b180 ***

  A work around is to set the shell environment variable MALLOC_CHECK_ to be 0 before calling gurgle.

  Tim has finished working on the web services project and has passed the deliverables to Stephen Gilmore for acceptance testing.

  Graham has finished setting up the postgresql server for the new teaching database server. He has developed a kx509 authenticated web page which students can use to request a database account and has added a toggle to allow the authentication to be either via kerberos or via a separate password file. In the latter case random initial passwords are generated by the web interface.

  Archie has finished building the project and .NET labs

  Condor is now ready to be rolled out to FC5 student lab machines (as soon as the new kernel with condor support reaches the stable release of DICE this Thursday).

  Rosemary has been working on a web page to allow staff to update the RAE 2008 data held in the Informatics database that refers to themselves. This work is nearing completion.

  A lecturer has requested a teaching package that includes a separate copy of Eclipse and Java so that the whole package is greater than 2GB in size. It can't currently be shipped as an rpm because the resulting rpm would be a single file with a size greater than 2GB which would require any software that was involved in its distribution to be able to handle large files. Although in this particular case it may be possible to find an alternative solution it does highlight a generic problem concerning the current limit to the size of rpms that we can handle.

  Tim reported that Maple is still not available on FC5. The licence arrangements have changed and Sheila is investigating.

• Services.

  The final AFS database server has now been installed in the JCMB server room. DNS A names will be created for afsdb1, afsdb2 and afsdb3 that point to static ip addresses. This information will then be forwarded to the AFS people to include in the self-serve db file which is distributed with AFS.

  Craig reminded us that by the end of this month all computing staff should be using an AFS home directory.

  A failed disk in the new SATABeast and a disk in the ATABoy, which was showing some errors, have been replaced.

  Neil has updated the vacation handling code so that when the vacation message is disabled the vacation cache is deleted.

  The amount of memory in thassos (a mirror server) has been increased; it had been labouring during some of the rsync processes.

  The global restriction on the Main wiki web has been removed so that now any DICE authenticated user can update any but a reserved set of pages (see Bugzilla 2325).

  A glitch last Friday with the new admin samba server is being investigated.

• User Support.

  Another 115 staff/PhD desktop machines have been upgraded to FC5 since the last meeting; another 351 machines of this type are left to be upgraded. Next week another 70 or so student lab. machines will be upgraded to FC5.

  Ken has moved all the unresolved bug reports that were in the groups that he previously managed to their new locations.

  Ken has finally finished the changes and testing of new rules and procedures to handle resit examination mark scaling for the ITO database.

  Morna has been working on improving the effectiveness of user documentation.

  Alison will be setting in train an RT modification to identify tickets that are unresolved and owned by nobody.

Appleton Tower refurbishment ramifications

George wished to draw people's attention to the work that was planned for Appleton Tower. The plans for levels 6, 7 and 8 were out for consultation. Professor Igor Goryanin and his research team will be decanted to the Darwin building prior to the work starting. Once the work on levels 6, 7 and 8 was complete then those on levels 2, 3 and 4 would be decanted to the new accommodation on levels 6, 7 and 8.

Alison mentioned the mock-up of a flip desk that is being considered for the refurbished levels 3 and 4 which is on display on level 5 in the north lab. People were encouraged to examine it and leave their comments.

Informatics Forum

Alastair has been requesting details of room occupancy and timing of moves so that we can start to plan the associated work that we must do.

The timing of the move into the Forum could potentially affect when it would be possible for staff to take annual leave next summer.

Although there will be no great pressure to quickly vacate Buccleuch Place and Forrest Hill the same can not be said about JCMB. We won't have to vacate the JCMB server room very quickly as there is no possibility financially of developing the area in the immediate future. Alastair will try to persuade College to keep the server area as a College machine room.

Alastair asked the User Support Unit to check that they didn't need to order any more desktop machines before next summer.

AOCB

1. Reserved UIDS

   Alastair raised the issue of an existing user having a uid in the reserved range. It appears that although this was noticed at the time all uids were rationalised, the work to convert that one was overlooked. The User Support Unit will arrange for the change to be finally made.

2. Registration of local account usernames

   The User Support Unit were asked to resurrect the documentation on creating local accounts and recording the details (in for example dice/options/localaccount.h and registering the accounts as functional accounts on the new IDMS) and place this information on the wiki.

3. Handling of an individual research database

   Julieta mentioned a research database that for historical reasons had been administered by herself. It should probably be passed to the Research and Teaching Unit; she will speak to Tim about this.