| Summary: | Jobs Data Reporting | ||
|---|---|---|---|
| Product: | Slurm | Reporter: | rl303f |
| Component: | Configuration | Assignee: | Brian Christiansen <brian> |
| Status: | RESOLVED INFOGIVEN | QA Contact: | |
| Severity: | 3 - Medium Impact | ||
| Priority: | --- | CC: | brian, da, rod, sfellini |
| Version: | 14.11.3 | ||
| Hardware: | Linux | ||
| OS: | Linux | ||
| Site: | NIH | Alineos Sites: | --- |
| Atos/Eviden Sites: | --- | Confidential Site: | --- |
| Coreweave sites: | --- | Cray Sites: | --- |
| DS9 clusters: | --- | HPCnow Sites: | --- |
| HPE Sites: | --- | IBM Sites: | --- |
| NOAA SIte: | --- | OCF Sites: | --- |
| Recursion Pharma Sites: | --- | SFW Sites: | --- |
| SNIC sites: | --- | Linux Distro: | --- |
| Machine Name: | CLE Version: | ||
| Version Fixed: | Target Release: | --- | |
| DevPrio: | --- | Emory-Cloud Sites: | --- |
1. There currently isn't a tool like tracejob. There is an open feature request for similar behavior -- Bug 980. To get the usage information like at the end of tracejob, you can use sacct. 2, 3. Currently there isn't a good way to do this. The possible workaround is to create a Prolog that gets the information from sacct and send it's own email. However, it's possible that the information might not be in the database yet. I have created a feature request, Bug 1536, for this request as we've discussed solutions to this issue in the past. Please reopen if you have anymore questions. Thanks, Brian err. I meant EpilogSlurmctld and not Prolog. |
Our transition to Slurm is from PBSPro which has some features that we find useful and would like to emulate within Slurm. 1) An invaluable tool for us under PBSPro is their 'tracejob' command which succinctly summarizes the entire history of a job. What is the slurm equivalent to 'tracejob'? If it does not yet exist, are there any plans for development? 2) At the end of a job, PBS includes information in the email to the user as shown: PBS Job Id: 8735160.cluster Job Name: test_job Execution terminated Exit_status=271 resources_used.cpupercent=398 resources_used.cput=00:00:50 resources_used.mem=109212kb resources_used.ncpus=2 resources_used.vmem=660060kb resources_used.walltime=00:01:31 How do we configure Slurm to include this job information in the email to user? 3) Similarly, what are best practices for configuring slurm to include resources used information in jobs' slurm.out file? Thank you!