Transitioning to Maui - Everything you need to Know

To make your transition experience (from Kupe to Maui) as seamless as possible, there are several important steps and points you need to take into consideration. Please read this entire article before trying to access Maui.

Note: Please report errors in the details provided here. As we will update the content on this page to correct errors or add new information do refresh this page from time to time.

Documentation & Training Information

We are in the process of completely refreshing the user support website, providing access to information categorised by:

  • Getting Started – all you need to know about the NeSI HPCs and how to apply for a Research Project Allocation, create an account on the new HPCs and login
  • Scientific Computing – all you need to know about how to run jobs, what software is available, links to training material, user guides and manuals, etc.
  • Storage – covering the file systems, I/O performance considerations, recovery of files, data transfer services, etc.
  • General – including important announcements and frequently asked questions
  • User Support – coming soon

If you have a problem – please look on the support website before raising a ticket. The search function works well.

Getting Started

Already have a NeSI Account

If you already have a NeSI account (and if required, a 2nd Factor Token), you are good to go, simply:

  1. If you are using two-factor authentication, simply login to lander02.nesi.org.nz, then
  2. ssh to login.maui.nesi.org.nz as you would have done on Kupe.

There is more detail here.  But if you are not able to login, then contact support@nesi.org.nz before doing anything else, such as a password reset.

New to NeSI

If you are new to NeSI – there are some initial steps that you must take (one-time only):

  1. Create a My NeSI Account and accept the Acceptable Use Terms and Conditions.
  2. Reset the temporary password to one of your choosing that conforms to the NeSI Password Policy. While special characters meet the password policy, we have seen problems with their use. It may be best to not use special characters.

If you must access NeSI from outside NIWA you will need to set up two-factor authentication.

If using two factor authentication, then you will be ready to log on to the NeSI lander node, then to jump to a Maui login node.

  • To log on to Maui, follow the instructions here.

Other Important Information

  • This Primer for Kupe Users Moving to Maui should help you understand some of the key differences between Kupe and Maui services.
  • This link provides useful information about the new HPCs.
  • You can gain a better understanding of the Maui filesystems here. In particular, please note:
    • Maui (shared with Mahuika), i.e. the HPCF, has a very large, very high performance scratch file system called /nesi/nobackup. You should do most of your computing in this file system.
    • Data in /nesi/nobackup will be automatically removed from time to time.
    • You will (in the near future) be able to use a “Librarian Service” to move data off disk and onto tape.
  • Details of how the workload manager (Slurm) is configured (for NeSI users) on Maui can be found here. In particular, please note:
    • We have introduced a nesi_debug quality of service (QoS) for rapid turnaround of test jobs. Each user may only have one job using the nesi_debug QoS setting at any time.
    • You will now need to specify the Slurm partition in which your job will run.
    • There is a Slurm primer here that provides additional information about how to submit jobs between the Maui and Maui_Ancil Slurm clusters.
  • You should have access to the same software you used Kupe. See how to find modules using the -S option on Maui (which will also be available of Maui_Ancil nodes in November).  Notice too that we are introducing a new support model.
  • The following link provides information on how to compile and link software on Maui:
  • If you run into a problem, please put in a ticket.

Significant Change to Software Stack

This is explained in more detail in the Primer for Kupe users moving to Maui. But the following describes the key changes. While all relevant software has been migrated from kupe, kupe_mp and the Virtual Labs, a few important changes have been made:

  • There is now a clearer separation between the maui_ancil, and Virtual Labs (CS500) and maui (XC50) software stacks: the maui software stack only contains software needed for large jobs (e.g., XIOS and the grib_api library), while general tools and libraries (e.g., Anaconda, NCO, and Mule) can now only be found on the maui_ancil and Virtual Lab machines. This ensures that we make best use of system capabilities, and avoid single core jobs running on maui (i.e. XC50).
  • The maui_ancil and Virtual Labs software stack now uses a more recent GNU toolchain based on GCC v7.1.0, to benefit from Intel Skylake architecture capabilities. Module names have therefore changed slightly with respect to Kupe.
  • To avoid clutter, older versions of the same software have not been migrated, with the exception of NCL.

Data Synchronisation

Unless you have advised us otherwise, all the data you have stored on Kupe’s  /home, /nesi/project, and /nesi/nobackup filesystems has been transferred to Maui’s /home, /nesi/project, and /nesi/nobackup filesystems. 

Syncing will continue until you delete the files:

  • /home/<user name>/KUPE_SYNC
  • /nesi/project/<project_id>/KUPE_SYNC
  • /nesi/nobackup/<project_id>/KUPE_SYNC

From either Kupe or Maui. 

Until you delete these files any file you create on Maui will be removed at the next sync.

Data Loss Risk Mitigation

The HPCF /home, /nesi/project and /nesi/nobackup filesystems were "snapshoted" at 6 pm on 19-Sep-2018.  Accordingly, should the final phase of syncing between Kupe and the HPCF delete files that you did not want deleted, they will be recoverable back to the state they were in at the time of these snapshots. This link explains how to recover deleted data from a snapshot of the /home filesystem on the HPCF.

NOTE WELL: If you do recover your data from a snapshot, you MUST  remove the KUPE_SYNC file - otherwise at the next sync your data will be deleted again.

Personal /nesi/nobackup Directories

Please recall the information we have previously provided re personal home directories on /nesi/nobackup. In particular that:

"On the HPCF your personal directory will be set to Read Only.  This will allow you to copy relevant data into the correct project directory (e.g. to /nesi/nobackup/niwa00111/bloggs), as it is not possible to support personal nobackup directories on the HPCF."

File System Quotas

Unlike on Kupe, filesystem capacity and inode quotas are being enforced on the HPCF.  To determine how much data you have, and what your current quotas are use:

nn_check_quota --help

Then re-run with the appropriate settings.

We will need to set new quotas for all users and projects transitioning from Kupe to Maui (i.e. the shared HPCF filesystems). You should review the purpose of each filesystem as explained here, and the associated permissions, then consider what quotas are needed to support your research project, then submit a request for a change in quota.

Time Frames

Synchronisation of some user's data in large projects and/or on the nobackup filesystem may not complete till after the 23/09.

For all other users, we ask you to stop data syncing (as indicated above) as soon as you are sure that you can continue your work on Maui / Maui_Ancil. Unless we encounter unexpected problems - this means that you should be able to complete your transition from Kupe to Maui by COB on 27/09.

This does not apply to NIWA Forecast Operations on Kupe.

Known Issues

As of 0900 on 08/10:

Data Synchronisation

  • No known issues.

 We will update this page as these tasks complete.

Cylc

  • No known issues

 

 

Labels: announcement
Was this article helpful?
0 out of 1 found this helpful