Operational Meeting

Apologies for absence

There were none.

minutes of the last meeting.

These were accepted.

Report from Computing Executive Group

This item was not taken.

Reports from units.

• Research and Teaching.

  Tim reported that they had fixed a bug in the default environment in that it was assuming that there was a local LDAP server. They also identified another bug in the roles command (part of the dice-accntutils rpm) which could not correctly handle multiple LDAP servers being named in the ldap.conf file.

  Condor is now ported to FC6 and there is an FC6 condor pool as well as an FC5 condor pool. Flocking from FC6 to FC5 has been turned off. It is possible in upgrading office machines from FC5 to FC6 some former FC5 office machines that were running condor are no longer doing so; this should be checked and rectified where appropriate.

  A long standing security hole in the way dbiproxy worked with ingres has been fixed locally.

• Infrastructure.

  George reported that levels 3 and 4 of the Appleton Tower had been handed over more or less on time and that the technicians had been working hard to complete the wiring of flip desks and patching at the switches. There had been a problem with the VoIP telephones when they were first used; when their traffic had been put onto a new VLAN there was no connection to the telephone exchange. However they are now using the original pilot service VLAN and the phones are now working.

  The previously reported problems with the air conditioning monitoring in the Appleton Tower server room appears to have been mainly prompted by what turned out to be a fault with one of the IS air conditioning units adjacent to their communications rack. However there was a fault that occurred with the air conditioning in our Appleton Tower server room this morning. George has reset all the units and reprogrammed one of the units. The temperature in the room was falling straight afterwards so that appears to have remedied the situation for the time being. The planned work for the Appleton Tower basement might encompass linking the air conditioning units up to the cooled water system. One of the UPSs has an ambient temperature sensor and George plans to eventually produce graphs from it available over the web. It could also be hooked into the new monitoring infrastructure.

  A second switched rack power distribution unit has been delivered and installed in the top of the new rack in the Appleton Tower server room.

  The new Appleton Tower routers (kubelik and jarvi) are now working and have taken over most of the functionality of the old ones (barenboim and jochum). The routing of wire E (the unmanaged UG/MSc VLAN) has been moved from JCMB down to the Appleton Tower. All the wireless access points have now had their firmware upgraded.

  Ian would be interested in any feedback on the Lantronix box. Neil and Stephen commented that it was not as convenient for analysing log information as the previous setup. [Tim left at this point]

  Toby would also be pleased to have feedback on the cosign facility, especially the documentation.

  George reported on his meeting with Sam Wilson to discuss the routing of traffic between Informatics hosts and the SRIF network. The agreed solution was that all Informatics hosts that need to talk to the SRIF network should have a second interface, either directly onto the SRIF network (implying no routing required) or onto a new VLAN of our own that would only be connected to the SRIF network via one of Sam's routers. The problem with the former is the lack of available network addresses, so the latter version of the solution may be more practical.

• Managed Platform.

  Stephen reported that as part of the new LCFG web site development he had produced two new perl modules (in lcfg-pkgtools and lcfg-pkgtools-perl). The qxpack utility had been completely rewritten and was now shipped as part of lcfg-pkgtools-perl instead of in lcfg-rpmcache. Advantages are that qxpack now works correctly on 64bit machines and on x86 machines it produces generally more accurate results.

  The new rpm slave for the central area, boreas, went into service on Tuesday.

  We are now properly supporting PAE (Physical Address Extension) kernels so that x86 DICE machines can now address 64GB of memory instead of the normal limit of 4GB (the host split has 8GB of memory and is running in 386 mode). The final bit of work was to add AFS support for the kernel.

  The LCFG slave mousa suffered an AFS glitch recently (the first of our machines to do so). It is however a known bug and there is a small patch available to correct it. Stephen will start to look into applying the patch some time after he returns from annual leave.

  It has been observed that some people have been directly editing defaults files on the lcfg master; this causes warning messages to be sent as backup copies of the files appear and disappear and unnecessary profile recompilations to take place. If the circumstances are such that a defaults file needs to be updated without using the more usual method of shipping it in a new rpm then it should be copied somewhere else, edited, and only then copied back.

  Scientific Linux 5 (SL5) will be installable on machines via PXE from this Friday. It provides inf level support but not dice level yet.

• Services.

  Craig reported that they have brought another linux file server (halcyon) into service for the use of HCRC (who purchased the equipment). It is located in JCMB and has 20TB of storage. It is a Dell PE 1950, and is attached to a SATABeast (satabeast2.inf), a disk array of 750GB IDE disks with a maximum capacity of 42 x 750GB (approximately 30TB).

  The new commodity content management system is now providing a pilot service. It holds the migration web site and ICSA wish to move their web site there also. The commercialisation group in the Appleton Tower are also considering making use of the service.

  They have been looking at using the shezhu room booking system for booking resources in the forum (having completed the work for the Appleton Tower room and resource booking). They are also continuing to assess the usability of the University calendar service for future room booking system.

  Cosign capability has been added to the main Informatics web site. Neil is proposing to provide cosign protection for all of the web site, but it will still be possible to use kx509 or file based authentication as well as cosign.

  The CUPS component was to be put into the testing release last Monday but it will now have to go in to that release before next Tuesday.

• User Support.

  Ken reported that contrary to what he had said at the last meeting it had been planned to have all machines installed in Appleton Tower levels 3 and 4 by the start of this week (and not last week as minuted), and that had essentially been achieved.

  About 153 lab machines and 98 office machines had been upgraded/installed with FC6 in the last three weeks. Ken presented the following figures for numbers of FC6 and FC5 machines:

  FC6				FC5
  All	Labs	Office	Other	All	Labs	Office	Other
  691	275	393	23	326	31	128	167

  Since the last Operational Meeting the User Support Unit had handled 307 new RT tickets (equivalent to about 20 per working day) and resolved 65% of them. There had been a total of 308 tickets (including both new and existing tickets) resolved over the same period.

  Ken reported some figures showing resolution rates against age of recent RT tickets:

  Ticket Age D	Percentage Resolved (or rejected)
  1 week < D < 2 weeks	76%
  2 weeks < D < 1 month	78%
  1 month < D < 3 months	91%

  The total number of accounts (not including temporary ones) which have an AFS home directory is now 308 (an increase of 117 in the last three weeks).

AOCB

There was none.

The next meeting will be on September 26th in JCMB 2511 and chaired by Ken Dawson.