Technical support

From Alliance Doc
Revision as of 15:23, 30 June 2020 by Diane27 (talk | contribs) (Marked this version for translation)
Jump to navigation Jump to search
Other languages:

Ask support[edit]

  • Before writing to us, consider checking first on the system status page and the Known issues page to see if the problem you're experiencing has already been reported. If you can't find the information you need on this wiki, send email to the address below that you think best matches your need.
  • A well written question (or a problem description) will likely result in a faster and more accurate assistance from our staff (see a support request example below).
  • An email "Something is wrong" or "Nothing works" will take a long time to resolve, because we will have to ask you to provide missing information (see Information required below). Please do not make our staff play detective.
  • In the subject of the email, include the system/cluster name and a few words of what may be wrong. For example, "Job 123456 fails to run on the Cedar cluster". A good subject line really helps to identify issues at a glance.
  • Your Compute Canada username may not be apparent from your email. We ask you to register all email addresses that you use to communicate with us in your CCDB account. This way, our ticketing system will be able to recognize you automatically.
  • Please do not request help on a different topic as a follow-up on an old email thread. Instead, start a brand new one to avoid re-opening an old ticket.
Email addresses

Information required[edit]

We are asking you to register all email addresses you use to communicate with Compute Canada in your CCDB account. You can add multiple secondary email addresses at https://ccdb.computecanada.ca/email.

To help us help you better, please include the following information in your support request:

  • Your full name
  • Compute Canada username
  • University or institution
  • Cluster name
  • Scheduler Job ID
  • Provide your job submission script (copy and paste, or attach)
  • Provide your complete output files where you see errors (copy and paste, or attach)
  • Provide commands that you were executing
  • Avoid sending screenshots or other large image attachments except when necessary - the plain text of your commands, job script etc. is usually more helpful
  • Identify the software (including the version) you were trying to use
  • When did the problem happen?
  • Please write clearly. Adhering to a formal writing style, following grammar rules, capitalizing properly, and dividing text into paragraphs makes it easier for us to understand your question.
  • If you want us to access, copy or edit your files, or inspect your account and possibly make changes there, say so explicitly in your email. For example, instead of attaching files to an email, you may indicate where they are located in your account and give us permission to access them. If you have already granted us permission via the CCDB interface to access your files, then you do not need to do it again in your support request.

Things to beware[edit]

  • Never send a password!
  • Maximum attachment size is 40 MB.

Support request example[edit]

To: support@computecanda.ca

Subject: Job 123456 gives errors on the CC Cedar cluster

Hello:

my name is Alice, user asmith. Today at 10:00 am MST, I submitted a job 123456 on the Cedar cluster. The Job script is located /my/job/script/path. I have not changed it since submitting my job. Since it is short I included it in the email below:

#!/bin/bash
#SBATCH --account=def-asmith-ab
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=16
#SBATCH --time=00:05:00
{ time mpiexec -n 1 ./sample1 ; } 2>out.time

A list of the following modules were loaded at the time follow:

[asmith@cedar5]$ module list
Currently Loaded Modules:
1) nixpkgs/16.09 (S) 5) intel/2016.4 (t)
2) icc/.2016.4.258 (H) 6) imkl/11.3.4.258 (math)
3) gcccore/.5.4.0 (H) 7) openmpi/2.1.1 (m)
4) ifort/.2016.4.258 (H) 8) StdEnv/2016.4 (S)

The job ran quickly and the myjob-123456.out and myjob-123456.err files were created. There was no output in the myjob-123456.out file but there was an message in the myjob-123456.err output

[asmith@cedar5 scheduling]$ cat myjob-123456.err
slurmstepd: error: *** JOB 123456 ON cdr692 CANCELLED AT 2018-09-06T15:19:16 DUE TO TIME LIMIT ***

Can you tell me how to fix this problem?

Access to your account[edit]

If you want us to access, copy or edit your files, or inspect your account and possibly make changes there, you should state so explicitly in your email (unless you have provided consent via the CCDB). For example, instead of attaching files to an email, you may tell where they are located in your account and give us written permission to access them.