slideshow 3


Early retirement of Manda-Mothra partition on May 21, 2019
Over the past weeks, we have been experiencing slowness and instability on the Orga GPFS filesystem. We are very much aware this has an impact on user activities. Due to this unforeseen situation, we are going to retire Manda and Mothra partition earlier than originally planned. All the Manda and Mothra (Dell AMD and Lenovo) nodes will be powered off on May 21, 2019. At the meantime, Bode partition will open to all users.

This will greatly reduce the load on the Orga file system and allow us to speed up data migration to the new filesystem, Hydra. Once sufficient data has been moved, we can begin moving the high-speed disks currently on the Mothra-Manda partition to the Chimera partition and resolve Hydra performance issues.

Users, please change your scripts accordingly and start using Chimera for your computation. We are sorry for the inconvenience of this earlier retirement. Our goal is to continue to provide you with a high availability GPFS file system and a quality production cluster.

External user account renewal is required quarterly by PI
From this year, all external user account has to be renewed quarterly by the respective PI. This policy is enforced in order to be compliant with IT's security rules. The first renewal will be on May 6, 2019.
External accounts will be locked unless a renewal request is sent to by external users' PI. If you have an active external account, please contact your Sinai PI for the renewal.

2019-2020 Resource Allocation Renewal is closed now
The submission period for new and renewed allocation requests for the 2019-2020 allocation period is now closed. Determination of allocation awards will be completed by 15 June and implemented beginning 1 July. Those who have not submitted requests to renew their 2019-2020 allocation will have their folders made read-only on July 2019.

Chimera is open to all user
The Chimera compute partition is open to all users on April 1, 2019.
Town Hall Slides:
Slides for training class “Introduction to Chimera”:
Minerva old partition will be retired on July 1st.

Chimera queue adjustments
Several minor changes are being made to the Chimera queues:
Normal queue has been expanded to 270 nodes and set as the default queue.
User-project association is now enforced on all queues and you will need a valid project (-p acc_xxx) for your job to run. Gold server is not being actively queried on Chimera to reduce job failures.
A default memory of 3G per processor core is enforced if no memory requirement is specified.

GPU node in Chimera partition
The interactive GPU node is accessible by:
$ bsub -q interactive -W 01:00 -n 1 -R v100 -R rusage[ngpus_excl_p=1] -Is /bin/bash
The GPU compute node is accessible by:
$ bsub -q gpu -n 1 -R rusage[ngpus_excl_p=1] -W 00:10 echo “Hello World”
$ bsub -q gpu -n 1 -R v100 -R rusage[ngpus_excl_p=1] -W 00:10 echo “Hello World”
We will slowly move the rest of the v100 GPU nodes to Chimera partition.

TSM files expire in 2019
The TSM files that were archived in 2013 will expire in 2019. A reminder email will be sent to users 3 months ahead. There is no simple command to extend the file expiration date. If you would like to keep these files, you will need to retrieve them to disk and archive again. If you would like to query other users’ archived files, please send in a ticket.

Amazon AWS transfers from Minerva
If you are copying large amounts of data to or from Amazon AWS, then please keep an upper limit of 100MB/s via your AWS CLI configuration. Minerva is a shared environment and aggressive uploads from some processes is bound to throttle performance for everyone else.

We do not backup user data. It is the users' responsibility to ensure data is backed up in accordance with their funding agency or private organization requirements. Refer to "Preserving Data: Who’s Got Your Back(up)?" for helpful hints in developing your individual backup strategy.

All Announcements/Reminders.

Theme by Danetsoft and Danang Probo Sayekti inspired by Maksimer