Installation from source on linux - process killed

[draganal28]

Information

• Qiskit Metal version - latest from the source as per 22.10.2024:
• Python version - the one that comes along as specified in environment.yml:
• Operating system - linux: Linux ThinkPad-X1-Carbon-2nd 5.19.0-46-generic Documenting code style choices (Owner: Jeremy) (Due: Feb 26) #47-Ubuntu SMP PREEMPT_DYNAMIC Fri Jun 16 13:30:11 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux:

What is the current behavior?

I follow recommended installation from source (Option 1, create new environment) but each time I try it (and I have been trying for the last three days), there is the same outcome:
dragana@ThinkPad-X1-Carbon-2nd:~/qiskit-workspace/workshop_SCQHW_fall_2024/qiskit-metal$ conda env create -n qiskitmetal -f environment.yml
Collecting package metadata (repodata.json): done
Solving environment: | Killed

(attached option_1_install_new_environment.png screnshot)

dmesg:
[ 2810.246062] oom-kill:constraint=CONSTRAINT_NONE,nodemask=(null),cpuset=/,mems_allowed=0,global_oom,task_memcg=/user.slice/user-1000.slice/[email protected]/app.slice/app-org.gnome.Terminal.slice/vte-spawn-8c6f265e-a98d-43cd-8cef-bb86c7f858fe.scope,task=conda-env,pid=6308,uid=1000
[ 2810.246093] Out of memory: Killed process 6308 (conda-env) total-vm:6556824kB, anon-rss:5371692kB, file-rss:0kB, shmem-rss:0kB, UID:1000 pgtables:12164kB oom_score_adj:200
[ 2810.257458] Purging GPU memory, 0 pages freed, 0 pages still pinned, 5440 pages left available.

(attached dmesg_1.png screnshot)

However, the memory does not seem to be that low:
dragana@ThinkPad-X1-Carbon-2nd:~/qiskit-workspace/workshop_SCQHW_fall_2024/qiskit-metal$ free
total used free shared buff/cache available
Mem: 7761456 2034620 4449856 412128 1276980 4986912
Swap: 2097148 1167976 929172

(attached memory_snapshot.png screenshot)

I am experienced with working with conda environments and installation of sw and I have never experienced a case like this.
I also tried creating a new clean environment then installing one by one package from environment.yml but at fourth the same kind of kill process happens (oom).

I have succeeded in installing qiskit-metal on macos by using the same commands.

Steps to reproduce the problem

git clone https://github.com/Qiskit/qiskit-metal.git
cd qiskit-metal
conda env create -n qiskitmetal -f environment.yml

What is the expected behavior?

It is expected to have new conda environment named qiskitmetal (in my case the name I gave to new environment) created.

Suggested solutions

I would suggest the team inspect how installation process is realised regarding depndencies as that it is the only thing I could think of which might cause OOM situation. Although it seems very unlikely to need that much memory. Can it be some recurrent link or circular link chain?

[zlatko-minev]

Sounds like something very specific to your Lennox environment. You could try the virtual environment set up instead of conda Sent from my iPhoneOn Oct 22, 2024, at 4:26 AM, Dragana ***@***.***> wrote: Information Qiskit Metal version - latest from the source as per 22.10.2024: Python version - the one that comes along as specified in environment.yml: Operating system - linux: Linux ThinkPad-X1-Carbon-2nd 5.19.0-46-generic #47-Ubuntu SMP PREEMPT_DYNAMIC Fri Jun 16 13:30:11 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux: What is the current behavior? I follow recommended installation from source (Option 1, create new environment) but each time I try it (and I have been trying for the last three days), there is the same outcome: ***@***.***:~/qiskit-workspace/workshop_SCQHW_fall_2024/qiskit-metal$ conda env create -n qiskitmetal -f environment.yml Collecting package metadata (repodata.json): done Solving environment: | Killed (attached option_1_install_new_environment.png screnshot) dmesg: [ 2810.246062] ***@***.***/app.slice/app-org.gnome.Terminal.slice/vte-spawn-8c6f265e-a98d-43cd-8cef-bb86c7f858fe.scope,task=conda-env,pid=6308,uid=1000 [ 2810.246093] Out of memory: Killed process 6308 (conda-env) total-vm:6556824kB, anon-rss:5371692kB, file-rss:0kB, shmem-rss:0kB, UID:1000 pgtables:12164kB oom_score_adj:200 [ 2810.257458] Purging GPU memory, 0 pages freed, 0 pages still pinned, 5440 pages left available. (attached dmesg_1.png screnshot) However, the memory does not seem to be that low: ***@***.***:~/qiskit-workspace/workshop_SCQHW_fall_2024/qiskit-metal$ free total used free shared buff/cache available Mem: 7761456 2034620 4449856 412128 1276980 4986912 Swap: 2097148 1167976 929172 (attached memory_snapshot.png screenshot) I am experienced with working with conda environments and installation of sw and I have never experienced a case like this. I also tried creating a new clean environment then installing one by one package from environment.yml but at fourth the same kind of kill process happens (oom). I have succeeded in installing qiskit-metal on macos by using the same commands. Steps to reproduce the problem git clone https://github.com/Qiskit/qiskit-metal.git cd qiskit-metal conda env create -n qiskitmetal -f environment.yml What is the expected behavior? It is expected to have new conda environment named qiskitmetal (in my case the name I gave to new environment) created. Suggested solutions I would suggest the team inspect how installation process is realised regarding depndencies as that it is the only thing I could think of which might cause OOM situation. Although it seems very unlikely to need that much memory. Can it be some recurrent link or circular link chain? —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[draganal28]

So, you'd think it is linux environment. I can give it a try by increasing oom score from 200 (visible from dmesg output) to 1000 and run the same command again. I have installation on macos, though, so I'll test it in a little while back on linux and if I find the cause for such behaviour, I'll share it.

Thank you for a quick response.