Pseudo-terminal will not be allocated because stdin is not a terminal. ******************************************************************************** * ⊹ ✧ _____ * * Welcome to ✧ ✧ ⊹ _____/..../___ ⊹ * * __ __ __ ____ ____ ______ ______ ____ / /..../ \ * * / // / / // __ \ / _//_ __// ____// __ \ / /..../___ \ . * * __ / // / / // /_/ / / / / / / __/ / /_/ / / /......../ \ ⊹ * * / /_/ // /_/ // ____/_/ / / / / /___ / _, _/ / /......../ \ * * \____/ \____//_/ /___/ /_/ /_____//_/ |_| / ‾‾‾‾/.../ \ * * . \ /.../__ / * * * JU Pioneer for \ /....../ / * * ⊹ \ /....../ / ✧ * * ✧ Innovative and Transformative * \ ‾‾‾/../ / * * * \_____/./______/ . * * Exascale Research * ⊹ // ⊹ * * ⊹ . ✧ / ⊹ * ******************************************************************************** System in pre-acceptance, expect performance fluctuations and other side effects ******************************************************************************** 2026-04-20T10:10:40+02:00 ## Temporary decrease on memory available per node ## Due to an issue observed on the client side of the parallel filesystem, the compute nodes in JUPITER will have GPFS tracing enabled to capture more data about that issue to find a lasting solution. As a result about 2GB of memory will be used for that tracing, which results in a slight decrease in memory available per node for application code. Once the issue is reproduced and the information is captured that tracing will be disabled again. We apologize for the inconvenience. ******************************************************************************** 2026-02-23T14:00:00+01:00 ## JUPITER expected availability ## The system is now available again in a pre-acceptance mode for all projects. The powerboard retrofit campaign is progressing well and should be finished by the end of this calendar week (by 2026-02-28). The limitations imposed by the slicing of the compute racks will be removed afterwards. We expect system stability to increase significantly. The availability of the system in the coming weeks strongly depends on the progress on the MDC and the implementation of its building management and automation system. Another period of downtime is required to prepare the MDC for final acceptance. The provider anticipated the second half of February and a one week downtime, realistically we expect it to happen in March. This downtime will include a unavailability of the ExaSTORE system, so a login to JUPITER is not possible in that timeframe in March. We will communicate the final dates once they are communicated and confirmed. We are aware of the implications this has for your research and apologise for any inconvenience caused. ******************************************************************************** System in pre-acceptance, expect performance fluctuations and other side effects ******************************************************************************** * Status information also at https://go.fzj.de/status-jupiter * ******************************************************************************** 2025-05-09T15:35+0200 Information - JUPITER System Status JUPITER is in a bring up state and operated in pre-acceptance mode. That implies instabilities and unexpected node states. -------------------------------------------------------------------------------- 2026-06-11T17:12+0200 Minor Incident - ExaFLASH in production ExaFLASH is now available on all nodes in JUPITER. This filesystem offers now a scratch space with higher bandwidth and IOPS as compared to ExaSTORE. As for all scratch filesystems, there is a retention time, in this case of 30 days. If your workload can benefit from more IO bandwidth or IOPS please adjust your jobs to leverage this new filesystem. Be also mindful that the total space is limited, so quotas are therefore more strict here. The mount point is /e/fscratch and the corresponding environment variables are $FSCRATCH_ -------------------------------------------------------------------------------- 2026-06-23T13:00+0200 Critical Incident - MDC downtime During the stated time period the room hosting the admin, login and storage nodes of JUPITER will be on maintenance for acceptance test of the data center. Unfortunately that implies that access to the system won't be possible during that period, and other systems mounting the JUPITER filesystems won't be able to access the data. Please take that into account for your work and the data stored on the system. We apologize for the inconvenience. -------------------------------------------------------------------------------- Starting on 2026-07-12T12:00+0200 Planned - Automatic cleanup activation on $SCRATCH Attention: The automatic cleanup will be activated on $SCRATCH after the MDC Maintenance (Jun 23 - Jul 12). Please note that the retention time for the $SCRATCH filesystem is currently 90 days, as documented. ******************************************************************************** +------------------------------------------------------------------------------+ | This node is in maintenance. Overall system state visible on the status page | +------------------------------------------------------------------------------+ Connection closed by 134.94.0.140 port 22