|
Frequently Asked Questions (FAQ)
General
Compute Cluster
General
How do I get help?
The best way to request help from GS-ITS is to send an e-mail to gsithelp [at] u.washington.edu. This will automatically create a request for you in our tracking system. You can also stop by our office or speak with us on the phone. Contact details are listed on our Getting Help page.
How do I get an account?
All account requests must come from designated account sponsors. Typically the account sponsor for a lab or group is the PI or manager. Whether you're part of the UW community or an outside collaborator, there's a few steps you'll need to complete before your account sponsor can request an account for you. These steps are detailed on our accounts page. There's a section for UW faculty, students, and staff, as well as a section for outside collaborators.
How and where do I print?
If you're using a GS-ITS managed system, any printers for which you have rights, should already be set up on your system. If you're missing a printer or have any other questions about printing, contact GS-ITS.
If you're using a personal or other unsupported system, please refer to our Printing page where you will find installation instructions for Mac OS X and Windows XP.
What departmental computing resources are available?
Shared departmental computing resources include 12 terabytes (TB) of online SAN connected storage and a high performance cluster with an aggregate 210 CPU cores. The department has a dedicated data center with redundant cooling, UPS/Generator backed power, gas fire suppression, seismically braced racks, and key-card controlled access. We have a departmental IT support team of 7 professionals. External access to resources is provided by a pair of SSH gateways, nexus1.gs.washington.edu and nexus2.gs.washington.edu. Departmental systems are backed up to tapes which are regularly shipped off-site for 3rd party vaulted storage. Tape backups are done using a Spectralogic T950 tape robot with 8 tape drives and a 400TB capacity.
What is the GS-ITS backup policy?
GS-ITS managed servers are backed up every day. Our standard backup policy is to retain data for one year. For more details about our backup policies and services, please refer to our Backups page.
What is GSDS?
GSDS stands for Genome Sciences Directory Services. GS-ITS provides a web application for account management to all users, as well as special applications for account sponsors. This system of web applications and the underlying directory is referred to as GSDS. GSDS is a single repository for all systems managed by GS-ITS.
What shared filesystems can I access?
The department maintains a shared directory for interlab file sharing. Files are deleted after three days and each user may only store up to 300GB of data. It's accessible by both Linux (/net/gs/vol3/shared/public) and Windows (\\net-gs\public) systems. Many labs also own group specific shared space. Check with someone in your group or with GS-ITS for details.
Who can help me with email or mail lists?
Email and mailist services are provided by UW Technology. UW Technology has detailed instructions for mail client setup and information about mail lists. GS-ITS can provide assistance with mail client configuration and mail list setup.
How do I send mail to all faculty/students/staff/...?
See the Mail Lists page to find the right one, then send to the appropriate address. If you would like to be on a department mailist, please contact gsadmin [at] u.washington.edu.
Are there any department licensed software packages available?
No. However, the University of Washington has a number of site licensed packages as well as a number of discount programs for purchasing software. Details on these programs can be found on the UWare site. If you have a valid license and just need install media, GS-ITS can loan you media for most common software packages.
Compute Cluster
What is the hardware?
The departmental cluster is made up of 23 Dell PowerEdge 1950 servers, each with dual Xeon E5345 processors and 16GB of RAM and an IBM blade centers with dual Xeon based blades. Altogether the cluster has 210 CPU cores. Standard home directories are available from the cluster.
What is the queuing system?
The departmental cluster, as well as lab specific clusters, use Sun Grid Engine 6.1. Information about using Grid Engine can be found on the Compute Cluster page.
How is the cluster queuing software configured for fair sharing?
Queued job priority is given to users who have historically used the least amount of CPU time on the cluster. The cluster's historical usage record degrades by half every 2 months such that recent jobs count more against a users' resource usage than do older jobs. It's important to note that the queuing system does NOT affect jobs that are already running. It's possible for one user to monopolize the cluster while other user's jobs wait in the queue. This imbalance happens when many jobs, owned by a single user, are submitted to an idle cluster. If other users then try and submit jobs they may find all available slots filled by jobs owned by a single user.
Why are some nodes in the cluster turned off or in state "au"?
The cluster uses a power management system to save energy. Nodes are turned on and off automatically based on demand. Even if the cluster queue is unused submitted jobs may remain queued for up to 7 minutes while more nodes are brought online.
|
