Please note that this report pertains to activity during July and August 2011.
The WTB atrium screen has been replaced and is using the new digital signage system. Screens are ready to be installed on all floors of the WTB atrium and are awaiting E&B fitting. New silent PCs for the existing MSI and JBC screens have been ordered.
We are evaluating replacement projectors to upgrade other CLS meetings rooms.
We plan to update the compute nodes by end Q3 2011, including upgrading RAM on nodes to 32GB, SGE to the latest version (Univa Grid Engine), Univa Unicloud for power and systems management, purchased M6348 so we can remove the separate GPFS network and simplify deployment networking. We also plan to cross mount the cluster, LMF and Nemo GPFS filesystems by end Q3 2011 and are evaluating options to use virtualised nodes instead of running web services on Sun V20z. By the end of the year we will have migrated the PostGres server to a virtual machine.
On 12 July we received indications of path thrashing in the cluster, reducing I/O performance. We tracked the problem to an errant cluster node and resolved the issue.
Our Apple Mac project to migrate all viable workstations to OS X 10.6 and decommission all non-Intel Macs is progressing, and we have a test environment using OS X 10.7 authenticated against Active Directory (LIFESCI-AD).
Our Windows 7 migration project is progressing division by division. We have completed migration for SLSL&T and where individual groups or PCs remain to be done, we have initiated migration for the next division; we will return to complete other divisions later as a mopping-up exercise. In parallel, we are also introducing additional Windows servers for divisions or research/support groups on request - new terminal servers are being developed for MAJF and CLS Finance groups.
We are evaluating a Remote Desktop Gateway for accessing new AD-authenticated terminal servers, which will streamline access to users connecting off-site and around campus. The 2X web portal project has been postponed pending the next version - initial tests didn't prove stable enough.
We will be evaluating VMWare View when version 5 is released, and one of the features this will offer is on-demand creation of virtual desktops based on pre-defined configurations.
We are progressing activity with existing Filemaker-based lab databases as well as restructuring administrative systems to better integrate with the University's data warehouse when this becomes a production system. We are also considering custom PHP interface development to improve access to scientific datasets, although this may not be possible unless additional resources can be made available within the team.
The University's project to move its email service to a cloud-based provider has agreed on Microsoft Office 365, which was launched late June 2011. Full details about this project are available from the project's blog at: http://blog.dundee.ac.uk/dundeemail/
The original plan to migrate during summer 2011 has been delayed and the plan now "is for staff and students to move at the same time, during the 4th quarter of 2011. Some areas may not be able to move for operational reasons - these accounts will still be able to send and receive emails etc as normal."
Note that as part of the email migration, mail archives will also be transferred to Office 365, but only if they are stored in a Groupwise format. For Life Sciences users, we have provisioned networked storage and this has been used by many users for several years. Information on how to setup your Groupwise archive is available at: http://www.lifesci.dundee.ac.uk/services/lsc/services/email/groupwise-archive
The University has also launched a drop box service as an alternative to sending large attachments through the email system. Details of this service are available at: https://dropbox.dundee.ac.uk/
On 18 July we noticed very slow performance from Groupwise and reported the problem to ICS. Performance returned to normal after ICS moved the service to another server, with a brief interruption to service around 17:17.
On 26 July ICS advised that a problem with part of Groupwise was affecting a large number of users, and service was restored for Life Sciences users by the afternoon.
On 16 August some Life Sciences accounts were affected by a failed Groupwise server and the problem was resolved by midday.
On 29 August mail addressed to recipients outside the University were not being sent, but flagged as Pending. In addition, on 30 August ICS scheduled a restart of the Groupwise cluster between 18:00-20:00 making the service unavailable to all users during this time.
On 31 August the Groupwise Webmail service went offline and the service was restored later in the morning.
We plan to complete upgrades, by end Q3 2011, to the server network using Dell M6348 stacked switches and deploying 10Gb Ethernet connections from these switches to the network core. Later in the year we will complete upgrades to access switches to 10Gb uplinks.
The Lifesci wireless network is now available in Carnelley and the Access Points (AP) are configured and mainained by LSC staff.
We are evaluating migration to IPv6 connectivity between CLS and UoD as well as transition from RFC1918 and NAT to real IP addresses. Both of these projects are dependent on ICS, and they are working to provide CLS with the necessary IP netblock. We have completed a test IPv6 deployment on capable equipment within the complex.
We aim to deploy Network Admission Control to manage guest access, and private VLANs with ACLs for all instrument/acquisition machines by the end of the year (Q4 2011).
Maintenance work on 25 July included upgrading wireless controllers to support new N-standard access points, although a fault rendered the wireless network service ofline until it was restored at 09:50.
ICS planned maintenance on Saturday 20 August to upgrade the telephony system. This was anticipated to reset Cisco telephones during the upgrade and was successfully completed by 10:40.
As part of our Windows 7 project, we have also developed a new password management tool to support the requirement for a new directory, LIFESCI-AD, as well as to help synchronise password changes between UoD and Life Sciences passwords. The best way to change your password is via the Life Sciences Directory website at: https://directory.lifesci.dundee.ac.uk/password/change
Most groups have shared networked printers that are available, and there are also several Canon multi-function copier/scanner devices on each floor in the complex. Printers connected directly to computers are not recommended or supported. In most cases, we are able to remotely configure connections to networked printers for your workstation.
Security of computers and data is very important. Our security policies are available at: http://www.lifesci.dundee.ac.uk/services/lsc/services/security/security-policies
Several projects are designed to improve and augment the existing security systems in place for Life Sciences computers and users: we will improve network security systems to support guest access while preserving security of research and teaching data; our password change system enables users to synchronise password changes between Life Sciences and University systems, but note that it does not apply to non-integrated systems such as Coda and Pecos; and our forthcoming remote desktop gateway will provide improved access to Windows terminal servers when users are working off-site or elsewhere on campus.
We will update Linux VMs to latest 5.5 via Spacewalk configuration management by the end of the year, and will also upgrade the VSphere environment to 5.0 by the same time. Our investigations into cloud management services is progressing and we have purchased Univa Unicloud.
On 1 July, one of our production fileserver clusters hung affecting DDU, Proteomics, sth-smb and jjb-smb network shares as well as LSC homedirectories. We recovered the filesystem, by restarting the locked processes.
Maintenance work on 12 July added processing capability and splunk monitoring to results.mascot, but we encountered problems that required a reboot during business hours.
On 18 August the PECOS procurement system was unavailable due to network routing problems at the hosting provider's location. This major incident continued throughout 19 August and was restored to normal operation by 16:48, although official confirmation was not received until 22 August. During 19 August, lab managers in Life Sciences reverted to using the Order Book as a contingency system, including orders to Stores.
All staff and postgraduate students in Life Sciences start with 5GB of personal network storage. This is accessible via several methods including over the web using WebDAV. Larger and faster network shares are available for purchase and usage is billed quarterly by debiting from users' cost centres.
Our new Alloy class storage service, which provides cost-effective storage in addition to existing Bronze, Silver and Gold storage classes (increasing cost and performance), is now available. We have migrated several network shares to Alloy storage following discussion with users. The iSCSI storage has been decommissioned and replaced with higher performance SVC, which provides storage virtualisation for improved resilience.
We are evaluating Tivoli Continuous Protection as a replacement for TSM backups, which can offer Windows laptops with more up-to-date (continuous) backups for critical data files.
Note that Windows 7 and Mac OS X both offer improved Offline File synchronisation services, and while these are available and can be activated on request, we do not recommend thee as an alternative to laptop backups.
Maintenance work on 25 July including moving ail-smb share to the new Alloy storage platform.
On Sunday 31 July one of the fileservers locked, affecting DMM, NMR, TOH, MAJF, SCILLS and DMPK groups. Attempts at a graceful restart failed so the server had to be repowered. The consequent filesystem consistency checks completed an hour later and the fileserver was back online.
We are migrating current Drupal-based websites to a distributed AEGIR Drupal system for better performance, management and development purposes.
The new Life Sciences Computing website uses this system and we welcome feedback on the design and content. We have integrated the IT support forms and ticket listings into the new website, so that you can report new requests and also follow our progress on your support tickets.
Information in the new website is public by default but there are confidential areas where you need to login, using your Life Sciences passwod, before you can view this information. Note that the IT reports are now also public, so will be visible when you are off-site without requiring login.
IT incidents and notices of planned maintenance will be posted to the home page, and are also copied to our new Twitter account, @lifescicomp, so please follow us if you want to be kept informed of issues.
