Here you'll find some of our most frequently encountered problems and recommended resolutions.
- Subscribing to BMRC mailing lists
- Resetting and creating new passwords
- Why can't I login via SSH?
- Do you support SSH Keys?
- Why did my qsub job fail?
We operate two mailing lists:
- email@example.com - for all users
- firstname.lastname@example.org - for GPU users
Both mailing lists are low traffic, high importance and are reserved for important service announcements from the BMRC team. All users are warmly encouraged to subscribe to our general mailing list and likewise GPU users are encouraged to join our GPU mailing list.
To subscribe to our general mailing list, send an email to email@example.com (from the address that you wish to subscribe - the subject and content are ignored so you may leave them empty) and then follow the instructions in the automated reply.
To subscribe to our GPU mailing list, please email your request to firstname.lastname@example.org.
If you have forgotten your password, please email us to request a password reset. You will receive a temporary password and instructions for how to set your own new password.
If you password has been reset, follow the instructions in the email (also copied below) to login and set your new a password.
- Login via ssh using your current password - the password change process will begin immediately. If you have received a password reset email, your current password is the temporary password in the email.
- When prompted for your Current Password re-enter your current password. If you have received a password reset email, your current password is (once again) the temporary password in the email.
- Only when prompted for a New Password should you enter your new password and then re-enter it when requested to Confirm New Password.
Your new password must contain at least 16 characters and contain at least three different character classes (lowercase, uppercase, numbers, symbols).
The password change is immediate and there is no need to log out and back in again. Contrariwise, if you are forcibly logged out after attempting to change your password, then your attempt was not successful and your password remains as it was.
If you cannot connect via ssh, look carefully at the error message received to help diagnose the issue and see what your options are.
|Message||Explanation & Resolution|
ssh: Could not resolve hostname rescomp1.well.ox.ac.uk: Name or service not known (likewise for rescomp2.well.ox.ac.uk)
The login gateways for the BMRC have changed to cluster1.bmrc.ox.ac.uk and cluster2.bmrc.ox.ac.uk on Tuesday 23rd November 2021. Please update your SSH configuration and discontinue use of the old names. Once logged in, these hosts will continue to report their local hostname as rescomp1-2.
ssh: connect to host cluster1.bmrc.ox.ac.uk port 22: No route to host (Likewise for cluster2)
All connections to our systems must originate from within the Oxford University IT network. For local users, you can connect either via ethernet or via an Oxford University-hosted VPN. For remote users, you will need to connect via Oxford University-hosted VPN. NB A VPN provided by another university or institution will not work.
Further details are available on the University's VPN page.
ssh: connect to host cluster1.bmrc.ox.ac.uk port 22: Connection refused
(Likewise for cluster2)
Your IP address has been temporarily blocked due to suspected malicious activity. This can happen e.g. if you or another user on the same IP address enter an incorrect password too many times. Please note that for blocking purposes, cluster1 and cluster2 are separate, so if you are blocked from one, you can still try the other.
If the block remains after ten minutes and you still cannot login, please Google "what's my ip address" and then email email@example.com with the address that Google reports. It will be in the format of four numbers (0-127) separated by dots e.g. 110.82.10.03
The BMRC cluster does not support login via SSH keys. UK research institutions are a target for hacking and in 2020 another UK research institution suffered a login breach that was due to SSH keys. As a result, many UK institutions no longer permit SSH keys. We are currently reviewing our login methods to ensure that they remain up to date with the latest security guidance.
Every job submitted via qsub receives a job id number - this is reported back to you when submitting the job.
The job id number is the most important piece of information in diagnosing a job. If you are contacting the BMRC team for help, please make sure to include your job id number.
If a computing job has failed (i.e. has ceased to run), try these steps in order:
- Check your log files, in particular the error log files. By default these files are named by combining your job script name, your job id and the letter 'o' for regular output or the letter 'e' for error output. For example, the script analyse.sh submitted as job id 2001524 would produce log files named analyse.o2001524 and analyse.e2001524. Checking the error log file will show if there are error messages that might help explain why your job failed. Some common problems are these:
Message in error log file Explanation & Resolution illegal instruction... You are running software built for one CPU architecture on another incompatible CPU type. The most common cause of this happening is the inclusion of the -V parameter to qsub, either on the command line or in your scripts.
We strongly recommend against using the -V parameter because it interferes with the normal process whereby our system automatically chooses the correct software for each system's CPU architecture. Instead, please include all necessary environment setup in your job submission script.
module command not found Please check that the first line of your script file is #!/bin/bash (with no empty lines preceding it). If you find that a script works when run interactively but produces a "module not found error" when submitted, this is likely the issue.
- If the log files have not helped, you can view the accounting information for this job by running qacct -j <job_id>. For example, with the example job mentioned above you would run qacct -j 2001524.
The accounting file shows lots of information. To begin with, focus on the values shown for: exit_status and failed. Some common problems are shown in the table:
Exit Status / Failed Message Explanation & Resolution
52 : cgroups enforced memory limit
Exit Status: 137. Failed: 100 : assumedly after job
It's likely that your job exceeded the maximum RAM allocation. Check the maxvmem value reported by qacct against the queue specification (in general, values over 15.2GB are exceeding the maximum).
You can either reduce the memory requirements of your script or increase the amount of RAM available to your job by selecting extra slots (see the section on parallel environments in the cluster guide).
Exit Status 137: Failed: 44 : execd enforced h_rt limit
Your job exceeded the maximum duration for its queue (see the queue time limits shown here). If you were using a short queue you could re-submit to the long queue to get extra time. Alternatively, you will need to try to shorten your run time so that your code completes in the time available - please contact us for advice. Exit Status: 0. Failed 27 : searching requested shell Your job submission script could not be read because it was prepared on Microsoft Windows and its text format is incompatible. Please run dos2unix <script.sh> on your script file to convert it to Linux format and then resubmit.