IPSY Docs
  1. Welcome
  2. /
  3. News

News

Known Issues

  • "Some" .hdf5 files remain locked after initial creation. The source of this problem is elusive, and makes little sense.

    The current work-around is to cp the file and then mv the copy over the original. This forces the file to be assigned a new inode, which invalidates the in-file locking. Madness.

Events

2020.07.02 - snake12 retired
The motherboard of snake12 died. Given the age of snake12 (purchased 2012), it has been decided to retire the machine. It lived a good life, and computed many scientifically relevant findings.
2020.02.05 - Major update to the docs
The majority of content from the INM-7 Docs has been ported over and adapted for IPSY. This represents over a year's worth of documentation fixes and improvements.
2020.02.04 - Dotfiles updated
The Dotfiles repo has been been entirely overhauled, porting fixes and improvements from the INM-7 dotfiles. It is located at /home/git/dotfiles.git on Medusa.
2019.11.10 - Degraded data array on Medusa
Zing (the data node) currently has a bad SSD drive in one of its arrays. The rebuild completed successfully with no data loss.
2019.08.01 - snake12 down
snake12 has a hardware failure and needs to be repaired.
2018.12.17 - ZFS upgrade and reboot; IPython history functional again

The data node's version of ZFS was upgraded to 0.7.12. This brings a wide range of fixes.

The cluster was rebooted for the update to take effect.

As a result of the reboot, the IPython history problem was cleared, though unfortunately not fully understood.

2018.12.14 - SPSS 25
SPSS 25 has been packaged for Debian. The university's SPSS 24 license has expired, so to continue using SPSS, you must upgrade to this new package. To upgrade, simply run apt-get update and apt-get install ipsy-spss25.
2018.08.31 - ZFS upgrade

The data node's version of ZFS was upgraded to 0.7.9. This brings a wide range of fixes.

The cluster was rebooted for the update to take effect.

2018.08.02 - Wine 3.0.1 installed cluster-wide
If you really need to shoe-horn your Windows-based workflow onto our Debian cluster, then there is a small ray of hope for you. If you can get your application to run via Wine 3.0.x, you can now run it across the entire cluster.
2018.07.09 - Increased number of Matlab toolbox licenses

The university has a new licensing agreement with Mathworks. In all practicality, there are now an unlimited number of Matlab and toolbox licenses (10,000).

Thus, Matlab users on Condor no longer need to limit the number of compute nodes used due to licensing constraints. The Condor/Matlab documentation has been updated to reflect this.

2018.07.09 - Fixed swap vs /tmp disk allocation on compute nodes

Due to a bug in the installation's preseed configuration, compute nodes with large hard drives were allocating the excess space to the swap partition rather than the /tmp directory. This has been fixed, and all nodes have been reinstalled.

It is now possible for jobs to run which need a large amount of local disk space rather than NFS.

2018.07.06 - Updated nibabel, nipype, indexed-gzip, fsleyes
Updated version of these packages have been installed, which should finally allow them all to coexist in fully updated harmony. Previously, many tools were displaying warnings, and a downgraded version of nibabel was needed to keep everything functional.
2018.06.13 - Signing Key for IPSY's Debian Repo Expired

The signing key for IPSY repository of Debian packages on Kumo expired. It has been updated and the updated key deployed to all cluster systems. If this is affecting you on your local system, run the following:

  • curl http://kumo.ovgu.de/debian/ipsy_apt.gpg.key | sudo apt-key add -
  • sudo apt-get update
  • sudo apt-get install --only-upgrade ipsy-keyring

If prompted about a conflicting ipsy.gpg file, respond with Y.

2018.04.11 - Condor jobs fail to start on snake4
When jobs attempted to run on snake4, they would bounce between running and idle and complain in the logs about a "Shadow Exception". The cause was a deeply mangled /etc/passwd file. The node has been reinstalled.
2018.04.05 - fsleyes crashes on start
An updated dependency of fsleyes caused it to crash. The bug was reported, the upstream maintainer released a fix, and that fix has now been deployed.
2018.03.16 - DataLad Upgrade
DataLad was upgraded and moved from a system package to a singularity container. Most users shouldn't notice a difference, but if you were using any of its Python libraries directly, they are no longer installed system-wide.
2018.03.13 - HeuDiConv/Nipype Fixed
Nipype was failing (prematurely), complaining about an outdated version of Pydot. Until the real fix is applied, an updated version of Pydot has been backported, which seems to resolve the problem.
2018.03.11 - ZFS upgrade

The data node's version of ZFS was upgraded to 0.7.6. This brings a wide range of fixes, especially performance related. Hopefully this will end the elusive "some files, take 1+ minute each to delete" problem.

The cluster was rebooted for the update to take effect.