Skip to content

Latest commit

 

History

History
240 lines (197 loc) · 12 KB

File metadata and controls

240 lines (197 loc) · 12 KB

Users gain access to a given cluster after starting a project in the Racimo group. Contact Fernando to obtain access.

racimo cluster

The "racimo" cluster is located in the "unicph" domain. To access these compute nodes, connect to the university's Cisco AnyConnect VPN.

Once you are connected to the vpn, you can ssh directly to a racimo cluster compute node. Available nodes are racimocomp<NN>fl, where <NN> is in the range 01--07, and racimogpu01fl.

racimogpu01fl

This node has 5x Tesla T4 GPUs, and in general should be used only for jobs that (also) need GPU resources. See gpu01.md for additional details about using the GPUs on this system.

cluster nodes

hostname CPU cores RAM GPU notes
racimocomp01fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp02fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp03fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp04fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp05fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp06fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimocomp07fl Xeon 6248R @ 3 Ghz 96 768 Gb
racimogpu01fl Xeon 6248 @ 2.5 Ghz 80 768 Gb 5x Tesla T4 See gpu01.md. No /scratch folder.

filesystems

Users have 10 Gb of storage under $HOME, which is enough for only small items. Temporary folders are not shared between cluster nodes and are not backed up. In general, other folders are both shared between cluster nodes and backed up.

path available space type shared between nodes description
$HOME 10 Gb network yes Personal directory.
/tmp 30 Gb local no Temporary storage.
/scratch 1 Tb local no Temporary storage.
/projects/racimolab/scratch 50 Tb network no Temporary storage.
/projects/racimolab/people/<kuid> 100 Tb network yes Create your own folder and use it for your projects.
/projects/racimolab/data 50 Tb network yes Use for your data.

dandy cluster

Racimo group members should use the racimo cluster, not the dandy cluster.

The "dandy" cluster is located in the "unicph" domain and nodes are accessed via the VPN just as for the "racimo" cluster.

Available nodes are dandycomp01fl and dandycomp02fl.

willerslev cluster

The willerslev cluster is shared with the GeoGenetics group, and should only be used for collaborative projects with that group. This cluster is accessed differently from the racimo cluster, and does not share filesystems in general (e.g. $HOME folders are shared within nodes of a cluster, but not shared between the two clusters). Using the racimo cluster should be preferred.

The "willerslev" cluster is located in the "science" domain, and it should not be necessary to connect to the VPN before logging in to this cluster. To access these compute nodes, first ssh to the head node ssh-snm-willerslev.science.ku.dk with your kuid and then ssh to the desired compute node (listed below). The latter step should not prompt for a password, and a password prompt at this step may indicate a transient problem or lack of access.

  • candy-snm
  • wonton-snm
  • dimsum-snm
  • taco-snm
  • biceps-snm
  • triceps-snm
  • compute<NN>-snm-willerslev, where <NN> is in the range 07--12.

Installing software

conda / mamba

Note: conda sucks. Use mamba instead. It sucks slightly less.

Software required to run a bioinformatics or analysis pipeline should be installed in a conda environment. It is likely that $HOME has insufficient space, so you should install conda somewhere else (e.g. /projects/racimolab/people/<kuid>/conda).

RHEL packages (rpm/yum)

NOTE: sudo privileges required. Ignore this section if you don't have sudo rights.

Suppose we wish to install emacs. Using the yum package manager, search for the package to be installed using yum list or yum search.

[srx907@racimocomp01fl ~]$ sudo yum list emacs\*
Updating Subscription Management repositories.
Red Hat Enterprise Linux 8 for x86_64 - Supplementary (RPMs)                                                     26 kB/s | 2.1 kB     00:00
EPEL 8                                                                                                           33 kB/s | 2.4 kB     00:00
Red Hat Enterprise Linux 8 for x86_64 - BaseOS (RPMs)                                                            31 kB/s | 2.4 kB     00:00
Red Hat Enterprise Linux 8 for x86_64 - AppStream (RPMs)                                                         40 kB/s | 2.8 kB     00:00
Red Hat CodeReady Linux Builder for RHEL 8 x86_64 (RPMs)                                                         37 kB/s | 2.8 kB     00:00
Red Hat Satellite Tools 6.9 for RHEL 8 x86_64 (RPMs)                                                             29 kB/s | 2.1 kB     00:00
Zabbix 5.2 RHEL 8                                                                                                27 kB/s | 2.1 kB     00:00
Installed Packages
emacs.x86_64                                                 1:26.1-5.el8                                      @rhel-8-for-x86_64-appstream-rpms
emacs-common.x86_64                                          1:26.1-5.el8                                      @rhel-8-for-x86_64-appstream-rpms
emacs-filesystem.noarch                                      1:26.1-5.el8                                      @anaconda
Available Packages
emacs-lucid.x86_64                                           1:26.1-5.el8                                      rhel-8-for-x86_64-appstream-rpms
emacs-nox.x86_64                                             1:26.1-5.el8                                      rhel-8-for-x86_64-appstream-rpms
emacs-terminal.noarch                                        1:26.1-5.el8                                      rhel-8-for-x86_64-appstream-rpms
emacs-yaml-mode.noarch                                       0.0.14-1.el8                                      Default_Organization_EPEL_EPEL_8
[srx907@racimocomp01fl ~]$ sudo yum search emacs
Updating Subscription Management repositories.
Last metadata expiration check: 0:00:08 ago on Wed 01 Sep 2021 02:16:12 PM CEST.
======================================================== Name & Summary Matched: emacs =========================================================
emacs.x86_64 : GNU Emacs text editor
emacs-common.x86_64 : Emacs common files
emacs-filesystem.noarch : Emacs filesystem layout
emacs-lucid.x86_64 : GNU Emacs text editor with LUCID toolkit X support
emacs-nox.x86_64 : GNU Emacs text editor without X support
emacs-terminal.noarch : A desktop menu item for GNU Emacs terminal.
emacs-yaml-mode.noarch : Major mode to edit YAML files for emacs
pinentry-emacs.x86_64 : Passphrase/PIN entry dialog based on emacs
xemacs.x86_64 : Different version of Emacs
xemacs-common.x86_64 : Lisp files and other common files for XEmacs
xemacs-devel.x86_64 : Development files for XEmacs
xemacs-filesystem.noarch : XEmacs filesystem layout
xemacs-info.noarch : XEmacs documentation in GNU texinfo format
xemacs-nox.x86_64 : Different version of Emacs built without X Windows support
xemacs-packages-base.noarch : Base lisp packages for XEmacs
xemacs-packages-base-el.noarch : Emacs lisp source files for the base lisp packages for XEmacs
xemacs-packages-extra.noarch : Collection of XEmacs lisp packages
xemacs-packages-extra-el.noarch : Emacs lisp source files for XEmacs packages collection
xemacs-packages-extra-info.noarch : XEmacs packages documentation in GNU texinfo format
xemacs-xft.x86_64 : Different version of Emacs built with Xft/fontconfig support
============================================================ Summary Matched: emacs ============================================================
ctags-etags.x86_64 : Exuberant Ctags for emacs tag format
e3.x86_64 : Text editor with key bindings similar to WordStar, Emacs, pico, nedit, or vi
mg.x86_64 : Tiny Emacs-like editor
vile.x86_64 : VI Like Emacs
xvile.x86_64 : VI Like Emacs

We can see from the above that the package name emacs is appropriate in this case. The package can then be installed with sudo yum install emacs. However, this would mean the package is only available on that specific compute node. For everyone's sanity, it is best to keep the installed packages synchronised across all the racimocomp nodes. So, simultaneous installation on all nodes should be preferred. E.g.

pssh -h ~/.ssh/racimocomp.txt -i sudo yum install -y emacs

Synchronising packages between redhat systems

Create a list of packages from a source node where the packages are already installed.

[srx907@racimocomp01fl ~]$ sudo yum list --installed > yum-list
[srx907@racimocomp01fl ~]$ head yum-list
Updating Subscription Management repositories.
Installed Packages
ImageMagick.x86_64                                  6.9.10.86-1.el8                                   @Default_Organization_EPEL_EPEL_8
ImageMagick-c++.x86_64                              6.9.10.86-1.el8                                   @Default_Organization_EPEL_EPEL_8
ImageMagick-c++-devel.x86_64                        6.9.10.86-1.el8                                   @Default_Organization_EPEL_EPEL_8
ImageMagick-devel.x86_64                            6.9.10.86-1.el8                                   @Default_Organization_EPEL_EPEL_8
ImageMagick-libs.x86_64                             6.9.10.86-1.el8                                   @Default_Organization_EPEL_EPEL_8
LibRaw.x86_64                                       0.19.5-3.el8                                      @rhel-8-for-x86_64-appstream-rpms
ModemManager-glib.x86_64                            1.10.8-4.el8                                      @rhel-8-for-x86_64-baseos-rpms
NetworkManager.x86_64                               1:1.32.10-4.el8                                   @rhel-8-for-x86_64-baseos-rpms

Double check the awk command to get the list of packages to install.

[srx907@racimogpu01fl ~]$ awk 'NR>2 {print $1}' yum-list | head
ImageMagick.x86_64
ImageMagick-c++.x86_64
ImageMagick-c++-devel.x86_64
ImageMagick-devel.x86_64
ImageMagick-libs.x86_64
LibRaw.x86_64
ModemManager-glib.x86_64
NetworkManager.x86_64
NetworkManager-config-server.noarch
NetworkManager-libnm.x86_64

Install those packages on the destination node where the packages need to be installed. It's possible that some packages won't be available on the destination node, e.g. due to differences in available repositories between the GPU and CPU nodes. These packages can be skipped by passing the --skip-broken flag to yum install.

[srx907@racimogpu01fl ~]$ awk 'NR>2 {print $1}' yum-list | xargs sudo yum install --skip-broken -y
...

Reporting Server Issues

Every now and then, some of these machines may have issues, either for a single user or for everyone. If after discussing the issue it is deemed necessary to report this to KU-IT, please follow these steps:

Step No 1

First log into the IT service portal and choose “create ticket” at https://serviceportal.ku.dk/HEAT/Modules/SelfService/#home

Step No2

Then select “Errors Reporting and Questions” from the Service Catalog Then click “Request”

Step No3

You should now have a new window open where you can report your issue. In the first text box, write a small summary of the issue you are experiencing. Remember to write the name of the machine e.g. racimocomp04fl in the "System name" box Choose “Error on a KU hosted HPC system” as the type of error from the drop down menu. Submit the ticket.

Step No3

You should now have recieved an automatic email in your KU mail about the ticket. If possible, once you have the ticket, feel free to forward the ticket email to any IT person that has recently contacted you about the server, informing them of the existence of the ticket and the urgency (if there is one). This can help as sometimes tickets may sit without a response for some time. If you do, CC Fernando in the mail.

Step No4

Patiently await the response from KU-IT!