Difference between revisions of "Frequently Asked Questions"
(21 intermediate revisions by 3 users not shown) | |||
Line 2: | Line 2: | ||
__TOC__ | __TOC__ | ||
+ | |||
+ | |||
+ | == '''Under Construction''' == | ||
+ | |||
== I Can't Access the Frontenac Cluster == | == I Can't Access the Frontenac Cluster == | ||
Line 10: | Line 14: | ||
== My jobs haven't started yet == | == My jobs haven't started yet == | ||
− | You can check your pending job by running the command squeue -u $USER. You queue time for your job depends on two major factors. First is the cluster load, you can check | + | You can check your pending job by running the command squeue -u $USER. You queue time for your job depends on two major factors. First is the cluster load, you can check the current load,i.e., running and pending jobs here: https://admin.cac.queensu.ca:4444/cgi-bin/cacMeters/cacMeter.cgi |
− | + | ||
− | + | ||
+ | Secondly, your priority might have dropped due to the recent usage of the compute resources from your group. You can check your group's fairshare level using sshare command (e.g., sshare -l -A def-hpcg9999). Your job priority depends on the LevelFS and anything less than 1 implies that you are over-served in the recent past and job are given a lower priority by the scheduler. This is to ensure fairshare usage amongst our user. Please note that your groups fairshare level will automatically recover with time, provide that the recent usage is low. | ||
== Can I use Conda for installing python packages == | == Can I use Conda for installing python packages == | ||
+ | We don't recommend installing python packages using conda on our clusters. We suggest you consider other options like a virtual environment or an Apptainer container, for the most complicated cases. You can find information on creating virtual environment here https://docs.computecanada.ca/wiki/Python/en#Creating_and_using_a_virtual_environment. Digital Research Alliance of Canada (DRAC) wheelhouse is available on our cluster, you can view currently available packages using command <tt>avail_wheels</tt> | ||
+ | |||
+ | <pre> | ||
+ | [hpcxxxx@caclogin04 ~]$ avail_wheels torch | ||
+ | name version build python arch | ||
+ | ------ --------- ------------- -------- ------ | ||
+ | torch 1.10.0 computecanada cp39 avx2 | ||
+ | torch 1.10.0 computecanada cp38 avx2 | ||
+ | torch 1.10.0 computecanada cp37 avx2 | ||
+ | </pre> | ||
+ | |||
+ | To view all available version of <tt>torch</tt> | ||
+ | |||
+ | <pre> | ||
+ | [hpcxxxx@caclogin04 ~]$ avail_wheels torch --all-version | ||
+ | name version build python arch | ||
+ | ------ ----------- ------------- -------- ------- | ||
+ | torch 1.10.0 computecanada cp39 avx2 | ||
+ | torch 1.10.0 computecanada cp38 avx2 | ||
+ | torch 1.10.0 computecanada cp37 avx2 | ||
+ | torch 1.9.1 computecanada cp39 generic | ||
+ | torch 1.9.1 computecanada cp38 generic | ||
+ | torch 1.9.1 computecanada cp37 generic | ||
+ | torch 1.9.0 computecanada cp39 generic | ||
+ | torch 1.9.0 computecanada cp38 generic | ||
+ | torch 1.9.0 computecanada cp37 generic | ||
+ | torch 1.9.0 computecanada cp36 generic | ||
+ | </pre> | ||
+ | |||
+ | If you require a specific version of python package that is not currently available on DRAC wheelhouse, contact our helpdesk at cac.help@queensu.ca. Please include the link for the source code. | ||
+ | == Installing R packages == | ||
+ | https://cac.queensu.ca/wiki/index.php/Software:R | ||
− | == | + | == My job doesn't seems run any faster on the cluster compared to my PC/Laptop == |
+ | tba |
Latest revision as of 20:28, 24 August 2023
Contents
Under Construction
I Can't Access the Frontenac Cluster
If you are receiving an 'Operation Timed out' message and you are from outside of Canada, you will need to use a VPN. If you are from the Queen's University Community, you can make use of the Fortinet VPN client offered by Queens.
I've Forgotten My CAC Password
Users can reset their CAC password their self, see https://login.cac.queensu.ca/pwr . Make sure you use the same email address as you used when you first signed up.
My jobs haven't started yet
You can check your pending job by running the command squeue -u $USER. You queue time for your job depends on two major factors. First is the cluster load, you can check the current load,i.e., running and pending jobs here: https://admin.cac.queensu.ca:4444/cgi-bin/cacMeters/cacMeter.cgi
Secondly, your priority might have dropped due to the recent usage of the compute resources from your group. You can check your group's fairshare level using sshare command (e.g., sshare -l -A def-hpcg9999). Your job priority depends on the LevelFS and anything less than 1 implies that you are over-served in the recent past and job are given a lower priority by the scheduler. This is to ensure fairshare usage amongst our user. Please note that your groups fairshare level will automatically recover with time, provide that the recent usage is low.
Can I use Conda for installing python packages
We don't recommend installing python packages using conda on our clusters. We suggest you consider other options like a virtual environment or an Apptainer container, for the most complicated cases. You can find information on creating virtual environment here https://docs.computecanada.ca/wiki/Python/en#Creating_and_using_a_virtual_environment. Digital Research Alliance of Canada (DRAC) wheelhouse is available on our cluster, you can view currently available packages using command avail_wheels
[hpcxxxx@caclogin04 ~]$ avail_wheels torch name version build python arch ------ --------- ------------- -------- ------ torch 1.10.0 computecanada cp39 avx2 torch 1.10.0 computecanada cp38 avx2 torch 1.10.0 computecanada cp37 avx2
To view all available version of torch
[hpcxxxx@caclogin04 ~]$ avail_wheels torch --all-version name version build python arch ------ ----------- ------------- -------- ------- torch 1.10.0 computecanada cp39 avx2 torch 1.10.0 computecanada cp38 avx2 torch 1.10.0 computecanada cp37 avx2 torch 1.9.1 computecanada cp39 generic torch 1.9.1 computecanada cp38 generic torch 1.9.1 computecanada cp37 generic torch 1.9.0 computecanada cp39 generic torch 1.9.0 computecanada cp38 generic torch 1.9.0 computecanada cp37 generic torch 1.9.0 computecanada cp36 generic
If you require a specific version of python package that is not currently available on DRAC wheelhouse, contact our helpdesk at cac.help@queensu.ca. Please include the link for the source code.
Installing R packages
https://cac.queensu.ca/wiki/index.php/Software:R
My job doesn't seems run any faster on the cluster compared to my PC/Laptop
tba