-
Notifications
You must be signed in to change notification settings - Fork 82
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Singularity missing? #68
Comments
Can you confirm that singularity is installed properly?
This suggests its not installed on your cluster and Nextflow can't call singularity properly. Maybe some missing module activation? You can by the way also simply run like this:
without requiring the |
I have rerun with the changes suggested above and on the other thread, I still get the 127 error:
|
Is singularity present on that machine? |
Yes!
|
Okay, it looks a bit weird because the error here:
Suggests via Also, this here:
states that you use a SLURM scheduler but didn't specify a profile for this. Do you use your own local configuration to submit jobs to a scheduler? In case so, I can add a configuration for your local hpc environment that should load required modules e.g. singularity automatically, as it could be the case it doesn't load singularity on cluster nodes right now. |
There’s no special configuration required to launch a slurm job on our HPC. Does NF require any special configuration?
Y
Cheers,
Yassine Souilmi
…________________________________
From: Alexander Peltzer <notifications@github.com>
Sent: Friday, November 2, 2018 11:53:19 PM
To: nf-core/eager
Cc: Yassine Souilmi; Mention
Subject: Re: [nf-core/eager] Singularity missing? (#68)
Okay, it looks a bit weird because the error here:
Process `fastqc (NGSD2_r2)` terminated with an error exit status (127)
Command executed:
fastqc -q NGSD2_r2_1.fastq.gz NGSD2_r2_2.fastq.gz
Command exit status:
127
Suggests via exit code = 127 that fastqc can't be found.
Also, this here:
Config Profile : singularity
=========================================
[warm up] executor > SLURM
states that you use a SLURM scheduler but didn't specify a profile for this. Do you use your own local configuration to submit jobs to a scheduler? In case so, I can add a configuration for your local hpc environment that should load required modules e.g. singularity automatically, as it could be the case it doesn't load singularity on cluster nodes right now.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub<#68 (comment)>, or mute the thread<https://github.com/notifications/unsubscribe-auth/ABr4IVBd7-pBmQR2c2S2Ptr9aIBQ3HL6ks5urEdHgaJpZM4YDlNE>.
|
Depends on the cluster quite a bit! Your error message could be explained by submitting jobs to slurm / cluster nodes, but without loading a Singularity module there and thus the tools inside the container cannot be resolved/used (error 127). How do you ensure that Singularity is up and running on the cluster? Do you load a module or something similar? |
Yes! I load the module and double check that it’s callable. Never faced an issue where a module needs to be loaded on the compute nodes.
I have been doing some digging, and the problem might be similar to (nf-core/rnaseq#29 (comment) <nf-core/rnaseq#29 (comment)>). But I am not really sure.
Thanks for your help.
… On 4 Nov 2018, at 8:27 pm, Alexander Peltzer ***@***.*** ***@***.***>> wrote:
Depends on the cluster quite a bit! Your error message could be explained by submitting jobs to slurm / cluster nodes, but without loading a Singularity module there and thus the tools inside the container cannot be resolved/used (error 127).
How do you ensure that Singularity is up and running on the cluster? Do you load a module or something similar?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub <#68 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABr4IVthTf9vtO6PrNkNmGvO8eWjJqCLks5urrnxgaJpZM4YDlNE>.
|
Hey!
|
The cluster is called Phoenix, and I load the module using the following command:
module load Singularity/2.5.2-GCC-5.4.0-2.26
… On 4 Nov 2018, at 8:33 pm, Alexander Peltzer ***@***.*** ***@***.***>> wrote:
Hey!
Can you let me know how you load the module and maybe just give me a name for the cluster? I can then come up with a -profile <your_cluster_name> that should run 👍
module load modules/singularity/2.5.2
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub <#68 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ABr4ITNGP7cNkVm8gX1dw-PxLr51TWF8ks5urruBgaJpZM4YDlNE>.
|
Thanks a bunch! I added a config profile for you to test - its called
|
Unfortunately, this didn't work. It first returned the following error:
Which I tried to circumvent by downloading the image manually, and than rerunning the workflow:
|
Ok, the first issue can be resolved by installing The second issue seems indeed to be related to the link you posted: nf-core/rnaseq#29 (comment) It looks like your cluster admin (?) didn't specify certain configuration parameters for singularity to access certain paths on your system. For example, we have a system where we can only run images and access files stored under I can help you in getting that up and communicate it appropriately if you want to, but we'd need to have a shared screen session / Skype to talk about this ... |
Sounds good to me, and thank you again for your support! |
No worries, happy to have some people testing the pipeline and giving feedback like this - thanks for testing and send me a DM/email if we should do a screen share or similar session to figure out whats going on. |
Heads up to document for future cases what goes on: Head node uses a different home than worker nodes on cluster, this leads to entire work directories not being readable, thus the error of missing things. Had similar issues on another cluster before, where the homes were linked as
|
Hi @apeltzer, I am the guy who installed and configured singularity on Phoenix which @yassineS is using. We use $SINGULARITY_BINDPATH to bind directory, see https://www.sylabs.io/guides/2.6/user-guide/bind_paths_and_mounts.html#binding-with-overlay for details This is my test script
This is what I got
The output above showed the binding worked well. However, nextflow still have the binding issue
I also ran another singularity image for a Computational Fluid Dynamics (CFD) application named OpenFOAM under /fast/users/a1222423/eager/work/8b/12a213d46743cf1ccdd3e71e23f23f with @yassineS account. The working directory is /fast/users/a1222423/eager/work/8b/12a213d46743cf1ccdd3e71e23f23f/pitzDaily and both input and output are under it. It worked well. All above indicates the issue is on nextflow side rather than singularity configuration, possibly the part integration with singularity or nextflow configuration (nextflow might redefine or not propagate $SINGULARITY_BINDPATH (?)), unless some bugs inside singularity itself Cheers, Bowen Edited history: Edited format and typos |
Aah, I see! I added the That could resolve this issue then, just make sure you have set your SINGULARITY_BINDPATH before running the pipeline then. |
Hi @apeltzer , I bring bad and good news I tested it with the new profile, it still had the same issue. I searched nextflow's document, and found below under images section on https://www.nextflow.io/docs/latest/singularity.html#images
Seems add |
Hi @uoabowen ! I added the required configuration to the acad-phoenix profile, so using the |
That's great! |
Originally posted by @yassineS in #67 (comment)
The text was updated successfully, but these errors were encountered: