Alison, Ross, Stephen
These were accepted.
Alastair will raise at CCPAG. Action deferred for 6 months
Done - now at end of netmon page
Amended to 'Move some reusable retired kit to AT server room' Deferred for 6 months while possibility of procuring additional rack space for AT is investigated.
Will be handled within Services Unit.
User Support still need to do this.
Pending meeting between RAT and Infrastructure Units. Deferred for one month.
Done, but code for this should be moved from individual hardware headers to server.h and small-server.h
Blog articles discussed:
Blog articles suggested:
Report from Computing Executive Group
Reports from units.
After some discussion, it was agreed that we should keep things as they are.
Most of the broken configurations have already been fixed.
Chris confirmed that a reboot is not necessary to introduce the new sleep features if the machine is already running the sleep component.
Toby asked about the status of this. OpenAFS 1.6.1 has just been tagged so Stephen hopes to start work on this next week.
Topics for discussion
It was agreed that removing the pam_kx509 module by default from the PAM stack on DICE machines seemed to be a good idea. MPU will investigate further
Neil asked whether we should have a lab worth of machines on the develop release to pick up errors. While it was felt that having an entire lab of machines on the develop release would be overdoing things, it might be useful to have a test lab develop machine.
It was agreed that NATing through the DICE box should prove sufficient for most needs and that this is what users should be advised to do. RAT and Mpu will collaborate on making sure this is documented. We will wait until a situation arises which NATing can't cope with before considering further solutions.
Alastair will respond to RT57024 which lead to this issue being raised in the first place.
The one outstanding issue appears to be that of user crontabs on machines not being backed up. Graham will write a blog article and investigate the issue further.
Dave has given permission for a trawl through user and group file space for cookies. Neil will now go ahead with this
Neil will add an extra column to his cookie list to identify which unit is responsible for it. All units to work through the list identifying which cookies they are responsible for.
All to make sure they have read the document before the next meeting
It was agreed that there had been several very useful talks at the conference. A question was raised about the current status of our USENIX membership. Alastair will investigate.
It was agreed that this could not be done until we have a better means of viewing and managing the queues on IS print servers. In the mean time, the Services Unit will liaise with IS to see if there is an alternative approach which can be taken.
It was agreed that the mpu would produce an apacheconf header file to allow connection limits to be switched on.
Ian raised the question of whether it made sense to stick with Apache for our web servers or whether it would be better to look elsewhere. No firm conclusion was reached though Graham noted that some form of supported lightweight httpd would be very nice.
Jennifer asked about Matlab licence usage, It would appear that the web page showing licence usage can be up to 5 minutes out of date and can thus show that there are licences available when in fact they have all been used up. RAT will look into the issue.
The next meeting will be on Wednesday 11th April 2012 in room 4.31
Please contact us with any comments or corrections.
Unless explicitly stated otherwise, all material is copyright The University of Edinburgh