MPU Meeting Tuesday 9th March 2010
AFS Component
Nothing happened, waiting for sign-off.
LCFG Server Refactoring
Nothing much happened.
Server Hardware
Stalled.
Installroot
Nothing much happened.
F12
Lots of work has been done on the package lists.
There were various problems related to the management of the F12 package repository. The worst of these was related to the fact that we ran out of space in a couple of directories stored in AFS with about 30000 files. The solution was to introduce a new sub-directory, named
hdrs
, into which the RPM headers are stored. This required changes to updaterpms and genhdfile to check for the existence of the new directory.
Stephen plans to get started on the F12/x86_64 port this week.
Alastair will talk to SEE to check how they want to be involved with this porting work.
Boot
This is mostly working now. Not much happened this week other than the code being packaged. Alastair will sort out the headers this week.
Miscellaneous Development
- Package Forge
- Stephen has spent about 1 week of effort putting together a prototype system which can build packages for users. He has something which mostly works as a proof of concept but he estimates it needs another week of effort to get it into usable shape. He will come back to this in April.
Project Prioritisation
The next round of project prioritisation is on Monday 22nd March. Stephen needs to add a project entry for remote desktop power management. Alastair needs to add a project entry for the DICE F12 work.
Operational
- amd usage
- The question was raised of whether we could remove amd usage from all the MPU servers. There are no obvious MPU services relying on amd/nfs being available so it should be possible to do this with no adverse effects.
- syslog logrotate file
- The syslog component removes the
etc/logrotate.d/syslog
when starting but upgrading the rsyslog packages puts it back. This bug is #233.
- Sleeping machines
- It looks like many of the lab machines are being forcibly shutdown with the power button by users who do not realise the machine isn't dead but is just resting. Chris is preparing a campaign of user education.
- cron logs
- The cron logs are all going into syslog. It would be better if this could go into
/var/lcfg/log/cron
instead. The cron log file will need the permissions changing.
- mrtg
- We don't need mrtg but it still generates lots of output. Alastair is planning to pull it from the DICE level, he will first check that no-one is using it.
- lcfg-syslog
- Needs a small fix to work with rsyslog on F12. Also needs the permissions on the syslog file tightening up.
- AT MPU servers
- The MPU servers mousa and split in AT are on the 212 wire. This is really a desktop wire so George would like us to move the servers to 202 addresses.
- lcfg-file
- Looks like running the file component from
boot.run
after updaterpms is not going to cause trouble. It should help cleanup small problems caused by dodgy packages. We need to watch that no-one abuses this facility though.
- MPU logging requirements
- We discussed the MPU logging requirements. Alastair will pass on the notes to George.
- Machine criticality
- We need to add some LCFG resources for defining machine criticality.
This Week
Alastair will:
- MPU disaster recovery
- Look at Panos's lcfg-network change
- MPU logging notes for George
- cron change for syslog
- Remove mrtg
- Talk to SEE about F12
- Add DICE/F12 project
- Boot component
Chris will:
- MPU disaster recovery
- F12
- Change the IP address for mousa
- User sleep education
Stephen will:
- MPU disaster recovery
- F12/x86_64
- Change the IP address for split
- Create project entry for remote desktop management
--
StephenQuinney - 17 Mar 2010