2006 User Survey Results: Services and Communications

• Legend
• Satisfaction with NERSC Services
• How Important are NERSC Services to You?
• How useful are these methods for keeping you informed?
• Are you well informed of changes?
• Comments about Services and Communications

Legend:

Satisfaction with NERSC Services

7=Very satisfied, 6=Mostly satisfied, 5=Somewhat satisfied, 4=Neutral, 3=Somewhat dissatisfied, 2=Mostly dissatisfied, 1=Very dissatisfied

How Important are NERSC Services to You?

3=Very important, 2=Somewhat important, 1=Not important

How useful are these methods for keeping you informed?

3=Very useful, 2=Somewhat useful, 1=Not useful

Are you well informed of changes?

Do you feel you are adequately informed about NERSC changes?

Are you aware of major changes at least one month in advance?

Are you aware of software changes at least seven days in advance?

Are you aware of planned outages 24 hours in advance?

Comments about Services and Communications:   25 responses

• MOTD / Communication of down times: 10 responses
• Satisfied: 5 responses
• Off Hours 24x7 Support: 4 responses
• Allocations issues: 3 responses
• Security issues: 2 responses

MOTD / Communication of down times:   10 responses

Frequently I lose contact with SEABORG, or SEABORG goes down, yet the MOTD says nothing.

sometimes when the pdsf cluster goes down, I'm unsure whether to report it or not. I usually assume someone else has becuase so many people use it. I find that the pdsf webpage, which is supposed to report its current status, is usually very lacking in staying updated on an hour-to-hour basis for these types of crises

The MOTD on PDSF is too cluttered, it's hard to extract any useful information even if it hasn't scrolled out yet.

The notices are sufficient to inform me.

The email updates work very well.

I would like better email communication of unplanned outages and downtimes.

There are perhaps too many e-mail messages - I tend to lose track of more important ones (like major outages) amongst the many I receive from NERSC.

I don't understand the comment about software changes. Is this with respect to computer libraries or routines or is this with regard to my program?

Satisfied:   5 responses

By in large, I am very impressed with the quality and professionalism of the NERSC staff and organization. In my 35+ years of running on academic and government computer systems, NERSC is best experience I have had.

I would like to express greatly gratitude for Dr. Andrew Rose's help on PDSF service.

All seems to be fine.

NERSC runs a first-class operation.

I am most satisfied with the NERSC supercomputing facility. I enjoy using this facility even from Canada and I am most grateful to DOE and my PI for providing me access to this state-of the art facility. Thanks .

Off hours 24x7 support:   4 responses

Off-hours 24x7 Computer and ESnet Operations support currently does not provide account support (e.g. password reset). It would be more helpful if it provide account support if off-hours as well in the future.

Off-hours support for PDSF is limited. The PDSF-staff are always helpful and responsive, but if they are not available (off-hours or on travel), critical issues sometimes get delayed. For non-critical issues this is fine, but for critical issue, such as GPFS usability (stale file handles) and filesystem slowness should be addressed by off-hours support staff.

I would like to see better off-hours support for PDSF. The system is currently being run with full support during business hours only, I would like to see this support to be expanded to off-hours.
The PDSF system has grown up to be a significant system in the NERSC infrastructure and deserves more attention during the non-business hours. One of the problems we occasionally encounter is that one of the batch nodes is "bad" (HW or SW malfunction) and that the node "eats" jobs. Jobs will start on the node, but immediately abort due to the problem on the compute node, then the following job starts, draining the queues down without actually completing the jobs. NERSC operators do not appear to be willing to fix these nodes by taking them out of the batch queue system or other approaches and a node like that can be malfunctioning for a whole weekend. There are some work-arounds for this problem, but they require fairly advanced knowledge of the batch system. I would like to see NERSC support PDSF in the off-hours.

Generally, PDSF personel seem to work very hard to respond to problems, even off-hours. Is off-hour support official, or just something they do to be nice? Really, PDSF should be officially supported 24/7 by NERSC.

Allocations issues:   3 responses

The Allocations process should make more use of external review, and be more open to researchers who are not currently funded by the DOE.

The automatic reduction in allocated hours is very inconvenient, because it does not let us schedule the runs as to suit the needs of the project.

Allocations process in terms of ERCP is good but the overall DOE philosophy of strongly favoring the big-big projects is somewhat short-sighted.
Many of us work on projects which are small now but may grow to be very major players --- i.e., from little acorns, might oaks grow and so on. Squeezing us out when small may lead to massive difficulties later.
One can argue it is best to develop MPP codes when a "small" player than when a major player (where bugs/mistakes could waste millions of CPU hours.

Security issues:   2 responses

I was not informed about the recent lock-out for more than a week. There was no announcement from the NERSC. I belive it is NERSC responsiblity to inform the users in advance of such a long recess. Have you ever thought that you wasted about 3% of users' annual research time? It is very precious, if you have not notied that yet.

As mentioned earlier, I was not informed when my password was deactivated.

Correction of the security issue that occurred last month was handled very professionally and efficiently.

Other

Overall, I am satisfied with NERSC services. But I have never yet heard a response to requests (placed indirectly through my PI) for an increase in the inode quota on my home directories, which would make my work much easier.