/dev/shm/__KMP_REGISTERED_LIBS is not cleaned up

[RandellPinegar]

A new file per process (access to the library) is created in /dev/shm which fills the shared memory causing the pods running in our kubernetes cluster to hang and fail. Is there a way to turn this file generation off or a mechanism to clean it up more regularly?

[VarshaS_Intel]

Hi,

Thanks for posting in Intel Communities.

Could you please let us know the OS details, Hardware details, Intel oneAPI Product, and version of the product you are using?

Could you please elaborate on your issue and let us know the complete process(or steps if any) you followed to reproduce the issue at our end?

Could you provide us with the complete error you are getting to investigate more at our end?

Thanks & Regards,

Varsha

[RandellPinegar]

We are using oneAPI 2021, 2022, and 2023.  We see the same results across all three.  We are on HP DL380G9 and DL380G10 servers on CentOS7, 8, and RHEL 7,8, and 9.  We see the same issues across all combinations although oneAPI2022/23 on RHEL9 seems to be the cleanest of the variants.  The main issue is that the /dev/shm/__KMP_REGISTERED_LIBS files are generated per instance of a call to the oneAPI library, and are then not in turn cleaned up after the process that called it is done.  These are then left over in the /dev/shm directory, which fills up causing our builds to fail.  Our application use is trying to build containers to be able to deploy within a docker environment which is why the /dev/shm filling up is critical, but we see this even in the environments that are not to run on docker.  Unfortunately, i cannot provide you with a copy of the files or directory as those are on a different network.

[VarshaS_Intel]

Hi,

We have not heard back from you. Could you please provide us with an update on your issue?

Thanks & Regards,

Varsha

[VarshaS_Intel]

Hi,

Thank you for your reply.

>>Unfortunately, i cannot provide you with a copy of the files or directory as those are on a different network.

For Community support, we would require you to submit a minimum reproduction sample code specific to your issue that provides us with the most relevant background information for triage.

In case you require privacy, and are unable to share the issue/sample with us publicly and if you are a licensed oneAPI product customer and/or member of Intel’s oneAPI Academic Program please submit a ticket for Priority support so that your application can be handled with the required data protection and privacy regulations.

Could you please specify and elaborate on the error also could you please provide us with the complete steps you followed which help us to reproduce the issue at our end?

Thanks & Regards,

Varsha

[VarshaS_Intel]

Hi,

We have not heard back from you. Could you please provide us with an update on on your issue?

Thanks & Regards,

Varsha

[VarshaS_Intel]

Hi,

We have not heard back from you. Could you please provide us with an update on your issue?

Thanks & Regards,

Varsha