342
edits
Changes
→Outage schedule
Jobs will not be scheduled with a run time (<code>--time=</code>) that extends into the beginning of a planned outage period. This is so the job will not be terminated prematurely when the system goes down.
* '''Oct 28-30 Argo''' will be offline for electrical power work, some upgrades of infrastructure machines, and some software and firmware updates. We anticipate return to service on Thursday October 31.** Note that at Argo we do not block new jobs from starting in advance of an outage. This is different from the practice at Siku and most Alliance clusters. We assume that all jobs There are protected by some form of checkpointing, and we will terminate any running jobs when the outage begins. If you want to submit a job in advance of the outage but do not want it to risk being terminated, you can use <code>--begin=2024-10-31</code> in your script or sbatch command to ensure that the job will only be considered for scheduling after the outagecurrently no planned outages.
<!--