Earlier this year we listed our 100th acknowledgement, 20 million computational hours and 2 million jobs processed since inception in 2009. If you have recently published work acknowledging usage of the HPC cluster please let us know.
PBS script errors:
We have noticed in the last few months that several researchers’ jobs were failing due to malformed job scripts. The problem was that the jobs start running, however due to the script errors the jobs run with default parameters and lack the series names. Hence the malformed jobs only have 1 core, a wall hour limit of 250 hours and interactive jobs will be terminated as the series name is missing. Some examples of malformed scripts are:
– placing a space between the # and the PBS directive, these are not comments and the line must start with #PBS, not # PBS
– placing “echo” statements before the #PBS directives.
– malformed shell interpreter line for example !/bin/bash instead of #!/bin/bash.
There will be a full data center shut-down on Sunday the 22nd of July at 09:00. The HPC cluster including all storage volumes will be shut down and all running jobs terminated. The cluster should be back up by 17:00.
One of our researchers, Jason Limberis, has written a tutorial on setting up script notifications using a telegram bot.