topic Re: Multiple video igestion containers not running in Edge Software Catalog

Multiple video igestion containers not running

Smrithi — Tue, 04 May 2021 01:58:28 GMT

We have to deploy at least 10 video ingestion modules in a single server machine. When we were trying to deploy 6 of these video ingestion modules, only 5 modules are running. The sixth one is always in the pending state and SS of it is attached. The SS of the module allocation logs are also attached. We tried reducing the memory constraints to 500 MB, but it was still the same. How to solve this issue?

Re: Multiple video igestion containers not running

Murni_J_Intel — Fri, 22 Jan 2021 03:52:16 GMT

Hi @Smrithi

This seems to be happening due to your current client node have insufficient resources for the extra container. if possible, you can increase the hardware resources and storage for it to accommodate the task.

Re: Multiple video igestion containers not running

Smrithi — Fri, 22 Jan 2021 05:19:16 GMT

When I tried to run 6 video ingestion modules in a setup having 8 cpu cores and 16 gb RAM, it was running fine and I was not facing any issues. But when I try to run the same setup in the server machine only 5 of these video ingestion containers are deployed. The configuration of the server machine I am deploying it in is 40gb RAM and it has 6 cpu cores. Does the number of cpu cores matter while deploying this?

Re: Multiple video igestion containers not running

Snehal_Yesugade — Fri, 22 Jan 2021 08:39:52 GMT

Intel team,

We would like to know if we need to do some configuration changes for optimal usage of available hardware; as changing hardware is not so easy.

Re: Multiple video igestion containers not running

Murni_J_Intel — Fri, 22 Jan 2021 09:06:45 GMT

Hi @Smrithi

Yes, it will be counted because it will be used to load all the running applications.

Re: Multiple video igestion containers not running

Murni_J_Intel — Fri, 22 Jan 2021 09:08:27 GMT

Hi @Snehal_Yesugade

May I get the current details of the device configuration and the current setting for your Video ingestion container?

Re: Multiple video igestion containers not running

Murni_J_Intel — Fri, 22 Jan 2021 09:40:42 GMT

Hi,

You can check on your current configuration of generated csl_app_spec.json file for the resources allocation for the Video Ingestion container.

The resources section describes the CPU and memory resources required by the Software Module. The sum of required resources of all Software Modules deployed on a node should not exceed the capacity of that specific node.

The value of the "CPU" attribute is in CPU milli units in the csl_app_spec.json. 1 CPU = 1000m.

You can refer to this information at [WORK_DIR]/orchestrator/docs/module.pdf in the resources section. You can modify this section for each of your VI containers to ensure it will not exceed the capacity of available CPUs on the node.

Re: Multiple video igestion containers not running

Smrithi — Fri, 22 Jan 2021 10:00:38 GMT

The configuration of the system we are using is that it has 40 GB RAM and 6 cpu cores. Is there any other specific system configuration that you would like to know? The value of the CPU attribute is 1000m for each 6 of the video ingestion containers, and the memory assigned to each of them is 500 MB in the resources field.

Re: Multiple video igestion containers not running

Snehal_Yesugade — Fri, 22 Jan 2021 12:02:45 GMT

Hi,

Resource section describes how much is allocated and not the required..........Am I correct?

If yes, then how to compute how much is required?

How much down can the configuration go?

Re: Multiple video igestion containers not running

Murni_J_Intel — Mon, 25 Jan 2021 03:14:40 GMT

Hi @Snehal_Yesugade

The problem with your setup that has been described previously was not able to run 6 VI containers on the same system. The error showing there is not enough CPU allocation for the requested deployment. In this case, you can manage the resource allocation for each of the deployed modules in csl_app_spec.json file as mentioned in the previous reply.

The total resources of your deployed module describe in the csl_app_spec.json should match your actual hardware resources in order to run all your module in the same system.

You can refer to the module.pdf file at [WORK_DIR]/orchestration directory.

as per the above example, you can change the CPU attributes for the Resources section to match your actual hardware resources.