Question: How do I register for BlueBEAR (HPC) account

Please visit  the BlueBEAR registration page and decide on a registration option on the web page from the following:

• Registering a Project
• Registering a username for the general BlueBEAR service
• Registering a username for specialist BEAR services     

Question:

Whom do I contact for support?

Please log a call through the IT Service Desk, using the "Make a request" option and choosing "Log a Fault with BEAR/CaStLeS".

Question: How do I change my password, I've forgotten it?

Please log a call with the IT Service Desk

Question:

How do I use the module to amend my environment

A module allows a user to dynamically change their environment to manage various applications and resources on BlueBEAR. More information can be found on the sourceforge website.

Question:

What projects and QOS can I use ?

Please use the command my_bluebear when logged into BlueBEAR

Question:

How many jobs can I submit ?

Please review the following website and go to section: "Resource Limits"

Question:

Can I use spaces (or other special characters) in my filenames ?

Some applications on the cluster, including the schedular, do not fully support spaces in filenames. File and directory names should, therefore, consist of letters and digits only. Note that, as with any Linux system, case is significant.

Question:

How can I see if my job is running?

Please use the "squeue" or "showq" commands.

Question:

Can I be notified when my job finishes?

Yes, add

#SBATCH --mail-type ALL

to your job submission script to get a message when each job begins and ends. See the help page for job submission.

Question:

How can I tell when my job is likely to start?

To get a rough estimate, please use the command "squeue"

Question:

My job has finished but I can't see the output files?

The RDS storage makes use of a filesystem caching technology. Files are first written to the cache area before being synchronised to the replicated RDS storage. Normally this synchronisation process completes within 1 - 2 minutes, however, if the cache is very busy it can take longer to complete.

The caching is protecting write performance of RDS for everyone, so the busier it is the longer it will take data to appear in RDS. This does not mean it's any slower at processing the jobs. If the cache was not there it would take longer and everyone using RDS would be impacted.

Question:

What applications are available?

You can find more information about an applications by following the application link page. The link on the application names provides documentation about the software and corresponding example scripts are made available to aid running a batch job

Question:

Why aren't there enough licences?

Please review the first paragraph from BlueBEAR Applications web page

Question:

I have an MPI problem, who can help ?

Unfortunately we currently do not have an expertise in this field, we would request the user to try and use the resources on BlueBEAR such as the Distributed Debugging Tool (DDT) and Optimization and Profiling Tool (OPT). Further information can be found at The University of Edinburgh website or consult the following link -Link for OPENMPI.

Question:

Why do I get the following error 'Argument list too long' when I type ' ls', rm, ls, cp or other commands in my directory?

This is due to the large quantity of files within the directory you are working in (Working dir.) and the maximum number of bytes a command can have is limited by the operating system, not the shell. The shell expands the ls argument and display an error which report that the command line argument exceeded the command line limit. Each command under Linux and UNIX-like systems accepts a parameter commonly known as command arguments (or args). For example, the command ‘cd /etc’ has one command line arguments, namely, /etc. Some command can accept more than two arguments. The average maximum length of the command line  for most Linux/Unix system argument is limited to “128K” within the Linux kernel. To find the limit for your system, type: getconf ARG_MAXlength. The best way to deal with this problem is a combination of the “find and xargs” commands. The syntax is:

find. -name | xargs ls

You can replace 'ls' with any other commands

Question:

What is the Exit_status I see in the email when a job ends?

All programs on Linux end with an exit status, also referred to as an exit code.  It is an integer in the range 0-255 that can be used to indicate success or failure of running a program.  In your job email this is the final exit status of your job, which is usually the status returned by the last command run from your job script.

An exit code of 0 is conventionally used to mean that a command ran without error (i.e. successfully completed) and any other number indicates an error occurred.  1 is usually used as a catch-all "unspecified error" code and other codes used to indicate specific error conditions.  You may find a list of exit codes and what they mean in the documentation of your program, especially if using commercial software packages.  For example the Linux Documentation Project has a pretty good table of reserved exit codes at http://www.tldp.org/LDP/abs/html/exitcodes.html and what they are used for in bash, although many are more generally applicable.

Question:

Why have I started to get the message "X11 connection rejected because of wrong authentication" ?

If you get this message when you start an X application, a possible reason is that your $HOME disk quota is full. Check using "bbquota". If you are low on disk space remove unnecessary files from your system:

Question:

Why do I get Invalid qos specification?

We use "QOS" as a way of controlling access to non-standard resources in BlueBEAR, for example bbgpu (the shared GPU  resources) or research group-owned resources. If you do submit to other QOS then you should note that the change we have made is that a QOS is now attached to a project, rather than a user as it was before. This means that when you submit to certain QOS, using the following in your job submission script:

#SBATCH --qos=[qos name]

You now must explicitly specify to use a project that has access to the QOS, using:

#SBATCH --account=[project name]

You can see a list of your QOS and projects using the command:

my_bluebear

When submitting a job, if you see an error message:

sbatch: error: Batch job submission failed: Invalid qos specification

then this means that the project you are using for this job is not allowed to use the specified QOS. In this case you should use the my_bluebear command to check which project you should use to be able to submit to the QOS.

If you would like any help getting your jobs to run then please log a ticket in the IT Service Desk using the "Other BEAR Request" option.