Frequently Asked Questions

From CC Doc
Revision as of 22:37, 14 March 2018 by Diane27 (talk | contribs)

Jump to: navigation, search
Other languages:
English • ‎français

Forgot my password

To reset your password for any Compute Canada national cluster, visit

Disk quota exceeded error on /project filesystems

Some users have seen this message or some similar quota error on their project folders. Other users have reported obscure failures while transferring files into their /project folder from another cluster. Many of the problems reported are due to bad file ownership.

Use diskusage_report to see if you are at or over your quota:

[ymartin@cedar5 ~]$ diskusage_report
                             Description                Space           # of files
                     Home (user ymartin)             345M/50G            9518/500k
                  Scratch (user ymartin)              93M/20T           6532/1000k
                 Project (group ymartin)          5472k/2048k            158/5000k
            Project (group/def-zrichard)            20k/1000G              4/5000k

The example above illustrates a frequent problem: /project for user ymartin contains too much data in files belonging to group ymartin. The data should instead be in files belonging to def-zrichard.

Note the two lines labelled Project.

  • Project (group ymartin) describes files belonging to group ymartin, which has the same name as the user. This user is the only member of this group, which has a very small quota (2048k).
  • Project (group def-zrichard) describes files belonging to a project group. Your account may be associated with one or more project groups, and they will typically have names like def-zrichard, rrg-someprof-ab, or rpp-someprof.

In this example, files have somehow been created belonging to group ymartin instead of group def-zrichard. This is neither the desired nor the expected behaviour.

By design, new files and directories in /project will normally be created belonging to a project group. The two main reasons why files may be associated with the wrong group are that

  • files were moved from /home to /project with the mvcommand; to avoid this, use cp instead;
  • files were transfered from another cluster using rsync or scp with an option to preserve the original group ownership. If you have a recurring problem with ownership, check the options you are using with your file transfer program.

For rsync you can use the following command to transfer a directory from a remote location to your project directory:

$ rsync -axvpH --no-g --no-p  remote_user@remote.system:remote/dir/path $HOME/project/$USER/

You can also compress the data to get a better transfer rate.

$ rsync -axvpH --no-g --no-p  --compress-level=5 remote_user@remote.system:remote/dir/path $HOME/project/$USER/

To see the project groups you may use, run the following command:

[name@server ~]$ stat -c %G $HOME/projects/*/

If you are the owner of the files, you can run the chgrp command to change their group ownership to the appropriate project group. To ask us to change the group owner for several users, contact technical support. You can also use the command chmod g+s <directory name> to ensure that files created in that directory will inherit the directory's group membership.

Another explanation

Each file in Linux belongs to a person and a group at the same time. By default, a file you create belongs to you, user username, and your group, named the same username. That is it is owned by username:username. Your group is created at the same time your account was created and you are the only user in that group.

This file ownership is good for your home directory and the scratch space, as shown hereː

                              Description                Space           # of files
                      Home (user username)              15G/53G             74k/500k
                   Scratch (user username)           1522G/100T            65k/1000k
                  Project (group username)            34G/2048k             330/2048
             Project (group def-professor)            28k/1000G               9/500k

The quota is set for these for a user username.

The other two lines are set for groups username and def-professor in Project space. It is not important what users own the files in that space, but the group the files belong to determines the quota limit.

You see, that files that are owned by username group (your default group) have very small limit in the project space, only 2MB, and you already have 34 GB of data that is owned by your group (your files). This is why you cannot write more data there. Because you are trying to place data there owned by a group that has very little allocation there.

The allocation for the group def-professor, your professor's group, on the other hand does not use almost any space and has 1 TB limit. The files that can be put there should have username:def-professor ownership.

Now, depending on how you copy you files, what software you use, that software either will respect the ownership of the directory and apply the correct group, or it may insist on retaining the ownership of the source data. In the latter case you will have a problem like you have now.

Most probably your original data belongs to username:username, properly, upon moving it, it should belong to username:def-professor, but you software probably insists on keeping the original ownership and this causes the problem.

If you already have data in your project directory with wrong ownership, you can correct this with commands:

$ cd project/$USER
$ chown -R username:def-professor data_dir

This will correct the ownership of the files inside data_dir directory in your project space.

Finding files with the wrong group ownership

You may find it difficult to identify files that are contributing to an over-quota condition in /project. The find command can be used in conjunction with readlink to solve this:

[name@server ~]$ lfs find $(readlink $HOME/projects/*) -group $USER

This will identify files belonging to the user's unique group, e.g. ymartin in the example shown earlier. If the output of quota indicates that a different group is over quota, use that group name instead of $USER.

See Project layout for further explanations.

sbatch: error: Batch job submission failed: Socket timed out on send/recv operation

You may see this message when the load on the Slurm manager or scheduler process is too high. We are working both to improve Slurm's tolerance of that and to identify and eliminate the sources of load spikes, but that is a long-term project. The best advice we have currently is to wait a minute or so. Then run squeue -u $USER and see if the job you were trying to submit appears: in some cases the error message is delivered even though the job was accepted by Slurm. If it doesn't appear, simply submit it again.

Why are my jobs taking so long to start?

You can see why your jobs are in the PD (pending) state by running the squeue -u <username> command on the cluster.

The (REASON) column typically has the values Resources or Priority.

  • Resourcesː The cluster is simply very busy and you will have to be patient or perhaps consider if you can submit a job that asks for fewer resources (e.g. CPUs/nodes, GPUs, memory, time).
  • Priorityː Your job is waiting to start due to its lower priority. This is because you and other members of your research group have been over-consuming your fair share of the cluster resources in the recent past, something you can track using the command sshare as explained in Job scheduling policies.

The LevelFS column gives you information about your over- or under-consumption of cluster resources: when LevelFS is greater than one, you are consuming fewer resources than your fair share, while if it is less than one you are consuming more. The more you overconsume resources, the closer the value gets to zero and the more your pending jobs decrease in priority. There is a memory effect to this calculation so the scheduler gradually "forgets" about any potential over- or under-consumption of resources from months past. Finally, note that the value of LevelFS is unique to the specific cluster.

How accurate is START_TIME in squeue output?

Start times shown by squeue are accurate but conditional.

Slurm schedules future START_TIME for high-priority pending jobs. These expected start times are computed from current information:

  • what resources will be made available by running jobs that complete; and
  • resources needed by other, higher-priority jobs waiting to run

The start times are conditional - Slurm invalidates these future plans if:

  • jobs end early, changing which resources become available;
  • prioritization is perturbed: submission of higher-priority jobs or cancellation of queued jobs

On Compute Canada general purpose clusters, new jobs are submitted about every five seconds, and 30-50% of jobs end early, so Slurm often discards and recomputes its future plans.

Most waiting jobs have a START_TIME of "N/A", which stands for "not available", meaning Slurm is not attempting to project a start time for them.

For jobs which are already running, the start time reported by squeue is perfectly accurate.