Frequently Asked Questions (FAQs)


Q. My submitted job is still in the queue - why is it not running?

A. There could be several reason why your job is not running:

  1. If you have access to the members.q queue, it could be that you and other users in your lab are currently using all your slots, which in case your jobs are being queued in the communal long.q queue instead.
  2. The queue where your job is sitting may be full. If so, your job will eventually run.
  3. You might have asked for compute resources that cannot be met, e.g. more memory or more cores than available on any compute node, e.g. -l mem=4048gb or -pe smp 256. If so, your job will never run. Either lower the job’s resource needs using qdel, or, alternatively, remove the job (qdel) and submit (qsub) a new one with adjusted resources.
  4. qstat -j <job_id> will provide details on why a particular job is not running. qstat -u '*' will show all jobs and their priority scores in the queue.

A: [For QB3 migrants] If you are coming from the QB3 cluster, please make sure that your job script does not specify any of the below QB3-specific SGE resources. A job submitted with one or more of these will sit in the queue forever.

Q. I tried to delete some jobs, and now they’re stuck in the state “dr”. How can I get rid of them?

A. The most likely cause of this is that node (or nodes) running your jobs crashed. Since the node(s) can’t report back to SGE and confirm the job deletion, the state of the jobs doesn’t change. To force the issue:

qdel -f $JOB_ID [-t $SGE_TASK_ID]


Q. I just started to get SSL-related errors when using qsub and qstat that I have never seen before;

error: commlib error: ssl connect error (SSL handshake error)
ssl error (the used certificate is expired)
unable to contact qmaster using port 6444 on host "q"

A. Your Wynton HPC account has expired. If so, you should already have received an email from us with instructions on how to request the renewal. If you have responded to that email, then it’s a mistake on our end (sorry) - please drop us another email.

Q. I cannot SSH into the development nodes - I get ‘IT IS POSSIBLE THAT SOMEONE IS DOING SOMETHING NASTY!’ and ‘Host key verification failed.’. What is going on?

A. This most likely happens because we have re-built the problematic development node resulting in its internal security keys having changed since you last access that machine. If the problem error looks like:

Someone could be eavesdropping on you right now (man-in-the-middle attack)!
It is also possible that a host key has just been changed.
The fingerprint for the ECDSA key sent by the remote host is
Please contact your system administrator.
Add correct host key in /wynton/home/bobson/alice/.ssh/known_hosts to get rid of this message.
Offending ECDSA key in /wynton/home/bobson/alice/.ssh/known_hosts:18
ECDSA host key for dev1 has changed and you have requested strict checking.
Host key verification failed.

then the solution is to remove that offending key from your personal ~/.ssh/known_hosts file. If we look at:

Offending ECDSA key in /wynton/home/bobson/alice/.ssh/known_hosts:18

we see that the problematic key is in this case on line 18. To remove that line, use:

$ sed -i '18d' .ssh/known_hosts

Then retry.

Files and folders

Q. Is it possible to have a common folder where our lab group members can share files and software?

A1. If you belong to a specific group, we can set up a /wynton/home/your_group/shared/ folder that group members (part of the same Unix group) have write access to. Any such files will count toward the disk quota of the user who owns the files. The typical use case is then that one or more members maintain subdirectories therein. If you need this, please drop us an email. Note, if the groups command reports lsd for you, then you do not belong to a specific group and can unfortunately not get a group-specific folder.

A2. Labs who purchase additional storage will get a /wynton/group/your_group/ folder. Files written in that folder will not count toward users disk quota.

Contributing to Wynton

Q. Our lab would contribute to Wynton HPC in order to increase our priority. How can we do this?

A. We welcome donations of any size. In return, your lab will receive a number of slots in member.q equivalent to the number of cores in a current Standard Node that your contribution would purchase. As of May 2020, that cost is $170 per slot.

Q. Our lab has some old nodes we’d like to contribute to Wynton HPC in return for priority. Will you take them?

A. Please get in touch with the Wynton team.