Kernel has died issue

[Alea4jacta6est]

Hi, I am trying to use your tutorials on the machine with 64RAM inside a docker container, but I get "Kernel has died" when I try to run this cell at the beginning of cell search tutorial:
from scimilarity.utils import lognorm_counts
from scimilarity import CellAnnotation, align_dataset

Do you have any idea why it happens?

[OmarSalem12]

Hi @Alea4jacta6est the dead Kernel is usually an indication of low memory. If you are running docker locally on your machine , it's important to make sure docker has no limit on memory usage. In docker desktop, go to settings > Resoruces, you will see how many CPUs / Memory docker is allowed to use. Make sure all 64 Gbs are usable by docker and try running SCimilarity again.

[Alea4jacta6est]

Thanks for a quick response, here is my /etc/docker/daemon.json:
{
"live-restore": true,
"log-opts": {
"tag": "{{.Name}}"
},
"storage-driver": "overlay2",
"mtu": 1460,
"default-ulimits": {
"memlock": {
"Name": "memlock",
"Hard": 68719476736,
"Soft": 68719476736
}
}
}
I use a GCP instance with 64GB, do you think it makes sense to increase memory to 128GB then?

[OmarSalem12]

The config looks right to me. The tutorial was tested with 64Gb of memory so I don't think 128Gb is necessary. It's also strange that it fails while importing methods. Loading the model is the memory intensive step. I'll consult with my colleagues and get back to you. Any logs that can help us troubleshoot where the issue is coming from will be appreciated. Thanks

[Alea4jacta6est]

It started to work on the machine with 80GB of memory, seems that 64 is just not enough