Suggestions for improving the official Textreport
Hello to everyone,
here are some suggestions from me to improve the current textreport.
General
- The job summary is missing and should be placed at the end of the text report.
- It should be possible to display the whole summary with a tail command
- The recommendations should be placed between the metric value tables and the job summary
- The "nodes" columns of different tables should have the same width and if possible other metrices of different tables should also be placed in such a way, that it does not seem "restless"
- One space character a the right hand side of every value in each column
- Put architecture data below Jobdata?
- IO: work, NFS and scratch
- If possible include ethernet, but it's not that much important
General information
- Inserting Jobname and Nodelist?
- Requested and Elapsed walltime in Slurm format(dd-hh:mm:ss) ?
CPU, memory and swap table
- CPU usage total should be made more comprehensible (average)
- CPU usage hwm --> CPU usage max?
- [/sec] --> [s]?
- Is there a summary for multi node jobs for every column? For example if there are 4 nodes then there can be the number of nodes inserted below thes nodes. Or the sum of all percentages of CPU total.
- Hyperthreading?
IO/Net
- Why is in IO no rate, just the sum of IO (read/write) over the whole job while in Network (Infiniband) the rates are given
Recommendations
- It should be explained in the text report, what the boundaries are, when a recommendations will be formulated for a special task.