News
News
HBSGrid Maintenance 8/7: Full Shutdown
- 03 AUG 2023
This is a reminder that our next maintenance is scheduled for Monday, August 7th, 8 am–12 pm ET. This will be a full shutdown for the HBSGrid compute cluster, MariaDB, and Globus data transfer nodes, for which ALL jobs and sessions will be terminated. Please plan appropriately.
...This is a reminder that our next maintenance is scheduled for Monday, August 7th, 8 am–12 pm ET. This will be a full shutdown for the HBSGrid compute cluster, MariaDB, and Globus data transfer nodes, for which ALL jobs and sessions will be terminated. Please plan appropriately.
During the maintenance window:
- You will not be able to log in to the HBSGrid via terminal or NoMachine.
- You will not be able to log in to the MariaDB database server.
- Globus transfers will be temporarily halted.
- All interactive and batch jobs, as well as database sessions, will be terminated.
What work is being performed and why?
- • We will apply security patches to all nodes in the cluster, to the MariaDB server, and to the Globus data transfer nodes. These changes ensure the safety and stability of our environment.
- • A storage array volume snapshot will be made for the MariaDB server.
- • We will patch a misconfiguration that allowed certain programs to run on the login nodes.
We thank you for your anticipated cooperation as we make improvements to our HBSGrid environment. Our next maintenance will occur on Monday, September 4th. If you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Maintenance 8/7: Full Cluster, MariaDB, & Globus shutdown
- 27 JUL 2023
Our planned work for August will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster, MariaDB, & Globus shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week. ...
Our planned work for August will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster, MariaDB, & Globus shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week.
We thank you for your cooperation and if you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Maintenance, 7/10/2023
- 06 JUL 2023
Our next maintenance is scheduled for Monday, July 10th, 8 am –12 pm ET.
...Our next maintenance is scheduled for Monday, July 10th, 8 am –12 pm ET. Access to login nodes, interactive/NoMachine sessions, and Globus nodes will be affected – please plan accordingly.
During the maintenance window:
- • You will not be able to log in to the HBSGrid via terminal or NoMachine.
- • All interactive jobs and NoMachine sessions (connected & disconnected) will be terminated.
- • Globus transfers will be temporarily halted.
What work is being performed and why?
- • We will apply security patches to all nodes in the cluster. These changes ensure the safety and stability of our environment.
- • We will update Globus with a patch for the connector to Google Storage.
- • We will add Lmod module files to central installation. This will enable users to access new applications before our upcoming Grid 3.0 release.
We thank you for your patience and cooperation as we make improvements to the HBSGrid. Our next maintenance will occur on Monday, August 7th. If you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Maintenance, Mon 6/5 – Full Shutdown
- 02 JUN 2023
This is a reminder that our next maintenance is scheduled for Monday, June 5th, 8 am–12 pm ET. This will be a full shutdown for the HBSGrid compute cluster, MariaDB, and Globus data transfer nodes, for which ALL jobs and sessions will be terminated. Please plan appropriately.
...This is a reminder that our next maintenance is scheduled for Monday, June 5th, 8 am–12 pm ET. This will be a full shutdown for the HBSGrid compute cluster, MariaDB, and Globus data transfer nodes, for which ALL jobs and sessions will be terminated. Please plan appropriately.
During the maintenance window:
- You will not be able to log in to the HBSGrid via terminal or NoMachine.
- You will not be able to log in to the MariaDB database server.
- Globus transfers will be temporarily halted.
- All interactive and batch jobs, as well as database sessions, will be terminated.
We are making two significant changes that will affect how you work. Please continue to read…
- • The first phase is to retire out-of-warranty hardware – a total of 7 compute nodes (~ 200 cores), most of which were not in production. Combined with the second phase, the net effect for the main queues will be a small increase in capacity.
- • The second phase is to reconfigure the queues to consolidate and prioritize the compute into a larger pool for general use. To accomplish this, we are removing the
mini
,micro
,test
, andvalidation
queues. The main queues (short_int
,long_int
,gpu_int
,sas_interactive
,short
,long
,gpu
,sas
) will remain unaffected.
To ensure your work continues without delay, it is imperative that, moving forward, you right-size your job resource requests: ensure that the cores and RAM you request match what the work requires. Please see our helpful guide on Choosing Resources. As a shared responsibility (for security and access) to yourself and your colleagues, we strongly encourage you to take the time to review your past resource requests and adjust these amounts when starting new jobs. Failure to do so will mean delays or failures in starting applications for yourself and others, and not being able to use particular resources; ultimately resulting in delays performing one's research.
What work is being performed and why?
- • We will apply security patches to all nodes in the cluster, to the MariaDB server, and to the Globus data transfer nodes. These changes ensure the safety and stability of our environment.
- • A storage array volume snapshot will be made for the MariaDB server.
- • We will remove 7 out-of-warranty nodes from the cluster (
nod3,4,5,7,8,10,11
). - • We will shift
nod25
into the main compute pool. - • We will remove the
test
,validation
,mini
,micro
queues. - • We will patch a misconfiguration that allowed certain programs to run on the login nodes.
We thank you for your anticipated cooperation as we make improvements to our HBSGrid environment. Our next maintenance will occur on Monday, July 10th. If you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Maintenance 6/5: Full Cluster, MariaDB, & Globus shutdown
- 23 MAY 2023
Our planned work for June will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster, MariaDB, & Globus shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week. ...
Our planned work for June will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster, MariaDB, & Globus shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week.
We thank you for your cooperation and if you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Maintenance, 5/1/2023
- 28 APR 2023
Our next maintenance is scheduled for Monday, May 1st, 8a –12p.
...Our next maintenance is scheduled for Monday, May 1st, 8a – 12p. Access to login nodes, interactive/NoMachine sessions, and Globus nodes will be affected – please plan accordingly.
During the maintenance window:
- • You will not be able to log in to the HBSGrid via terminal or NoMachine.
- • All interactive jobs and NoMachine sessions (connected & disconnected) will be terminated.
- • Globus transfers will be temporarily halted.
What work is being performed and why?
- • We will apply security patches to all nodes in the cluster. These changes ensure the safety and stability of our environment.
- • A storage array volume snapshot will be made for the MariaDB server.
- • We will change the cluster configuration so that GPU interactive sessions run in the gpu_int queue only. This will enable us to more closely monitor any problems with PENDing interactive sessions across the cluster.
- • We will change the MATLAB default version to 2022b from 2021b. This update was overlooked during the recent Grid3 update. Please see below if you wish to continue to use 2021b.
- • We will fix a bug that would permit jobs with small, specific resource requests to bypass the scheduler. This fix enables easier troubleshooting and full transparency for resources used on the cluster.
- • We will add a warning message to terminal sessions that are allocated on the login nodes and not the compute nodes. This should help make users aware not to run CPU- or RAM-intensive work on the login nodes, ensuring a better user experience for everyone.
Please note the following important changes:
- • After the maintenance, MATLAB 2022b will be the default version of MATLAB for the current Grid3 release (2022.11). If you wish to remain on 2021b, please select the 2022.01 environment when using the MATLAB launcher (GUI) or execute
module load rcs/rcs_2022.01
before starting MATLAB sessions via the terminal. - • Starting Monday, all GPU interactive sessions should be sent to the
gpu_int
queue instead ofgpu
. If you are using the Grid3 UI in NoMachine, you do not need to do anything. If starting interactive jobs via the terminal, use-q gpu_int
as the queue option – your interactive job will not be accepted if submitted with-q gpu
.
We thank you for your patience and cooperation as we make improvements to the HBSGrid. Our next maintenance will occur on Monday, June 5th. If you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Environment Changes for 4/3/23
- 30 MAR 2023
Our next maintenance is scheduled for Monday, April 3, 8 AM–12 PM. This will be a full cluster & MariaDB shutdown, for which ALL jobs and sessions will be terminated. Please plan appropriately. ...
Our next maintenance is scheduled for Monday, April 3, 8 AM–12 PM. This will be a full cluster & MariaDB shutdown, for which ALL jobs and sessions will be terminated. Please plan appropriately.
During the maintenance window:
- You will not be able to log in to the HBSGrid via terminal or NoMachine.
- You will not be able to log in to the MariaDB database server.
- Globus transfers will be temporarily halted.
- All interactive and batch jobs, as well as database sessions, will be terminated.
What work is being performed and why?
- • We will apply security patches to all nodes in the cluster. These changes ensure the safety and stability of our environment.
- • A storage array volume snapshot will be made for the MariaDB server.
- • We will apply FairShare to the
gpu
andgpu_int
queues. This will prevent users from hoarding GPU resources and allow users to get on the GPU node more equitably. - • We will enforce
gpu_int
usage. Interactive jobs submitted to thegpu
queue will be rejected; they must usegpu_int
. This will allow us to better monitor for PENDing jobs in thegpu_int
queue. - • We will release version 2022.11 of Grid 3.0. This ensures that researchers have access to the latest features; SEE BELOW for more information, as ACTION MAY BE REQUIRED on your part.
- • StatTransfer v16 will be installed, and bug fixes for Stata v17 will be applied. Please see below for more information on StatTransfer.
Grid 3.0 Release 2022.11
We are pleased to release an update to Grid 3.0 that will ensure that researchers have access to recent software releases and features. Please see our HBSGrid documentation for detailed information about the new software and versions available.
Python and R Users: If you have custom packages or modules for Python or R installed in your home directory, you will need to update them in order to continue using them. If these are already offered centrally, please consider using the central installation instead. If you need assistance or have any questions, please contact us at research@hbs.edu.
Using StatTransfer 16
StatTransfer 16 will be available on the HBSGrid cluster after the maintenance. We will release an update to Grid 3.0 shortly to include this. If you wish to start using it right away, you may access it via terminal, either in NoMachine or a shell session.We thank you for your anticipated cooperation as we make improvements to the HBSGrid. Our next maintenance will occur on Monday, May 1. If you have any questions/comments, please email us at research@hbs.edu.
Upgraded HBSGrid Account and Project Space Provisioning System Coming Soon
- 24 MAR 2023
RCS, along with our IT partners, are excited to announce the upcoming replacement of our HBSGrid account and project space provisioning system effective April 10, 2023. This upgrade will significantly improve the speed, efficiency, and automation of creating and maintaining HBSGrid accounts and project spaces. To successfully migrate to the new system, RCS will be unable to fulfill any requests for new HBSGrid accounts or project folders, or make changes to existing project folders (e.g., add additional people or storage space) between April 3 and April 7; if you anticipate that you or a guest user will need a new HBSGrid account or new project space, please submit your request by noon on March 30th. ...
RCS, along with our IT partners, are excited to announce the upcoming replacement of our HBSGrid account and project space provisioning system effective April 10, 2023. This upgrade will significantly improve the speed, efficiency, and automation of creating and maintaining HBSGrid accounts and project spaces. Features include:
- New and improved forms to request accounts and project spaces
- Streamlined guest HBSGrid account creation
- Automated emails to approve account and project space creation/maintenance
You can learn more details about what to expect from the new system in our HBSGrid documentation.
To successfully migrate existing accounts and project folders to the new system, RCS will be unable to fulfill any requests for new HBSGrid accounts or project folders, or make changes to existing project folders (e.g., add additional people or storage space) between April 3 and April 7. Your ability to use existing HBSGrid accounts and project spaces will be uninterrupted during this time.
If you anticipate that you or a guest user will need a new HBSGrid account or new project space, please submit your request by noon on March 30th. While you may submit requests during the outage, they will not be fulfilled until the system is back online.
What will change for users on April 10th?
New and Improved Request FormsWe will be retiring our old account and project space request forms. Instead, you will use new forms to request new spaces and accounts. Note that if you typically access forms from the Online Requests tab of our website, you can continue to do so; when the new system is live, we will update these links.
Streamlined Approval EmailsHBSGrid guest accounts (i.e., accounts for users who do not have HBS credentials) and all HBSGrid project spaces must have an approved HBS sponsor, usually a faculty member. The new system makes it easier than ever for sponsors to approve account requests and project space members by simply clicking a link in an email.
Please visit our HBSGrid documentation website for more details about how this system change might affect you. We thank you for your cooperation, and please reach out to RCS with any questions or concerns.HBSGrid Maintenance 4/3: Full Cluster & MariaDB shutdown
- 20 MAR 2023
Our planned work for April will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster & MariaDB shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week. ...
Our planned work for April will include security patches and updates of all machines in our environment, requiring system restarts. Accordingly, this will be a full cluster & MariaDB shutdown, and all interactive and batch jobs will be killed – please plan appropriately. We will send more details about the maintenance next week.
We thank you for your cooperation and if you have any questions/comments, please email us at research@hbs.edu.
HBSGrid Environment Changes for 3/13/23
- 10 MAR 2023
Our next maintenance is scheduled for Monday, March 13, 8 AM–12 PM. ...
Our next maintenance is scheduled for Monday, March 13, 8 AM–12 PM.
During the maintenance window:
- You will not be able to log in to the HBSGrid via terminal or NoMachine.
- All interactive jobs and disconnected NoMachine sessions will be terminated.
What work is being performed and why?
- We will apply security patches to login, Globus, and queue nodes. These changes ensure the safety and stability of our environment.
- We will apply bug fixes to address problems with LSF when using GPUs. This will fix some bugs and ensure that we have access to continuing support.
We thank you for your anticipated cooperation as we make improvements to the HBSGrid. Our next maintenance will occur on Monday, April 3. If you have any questions/comments, please email us at research@hbs.edu.
HBSGrid March Maintenance Delayed
- 03 MAR 2023
MariaDB Server Upgrade Delayed
- 01 MAR 2023