Process with multiple nodes at the same time

[Marcuzzo77]

Hi,

I have set all the nodes with the command:
docker run --rm -p 3000:3000 opendronemap/nodeodm

in webodm I see them connected and I can choose the node for processing (manual).

I wonder if it was possible to use them all together to speed up the process?
Marco

[Saijin_Naib]

Are there multiple machines you wish to process on? If so, have you deployed and configured ClusterODM as the job marshaller and deployed NodeODM to each machine?

[Marcuzzo77]

Hi,

yes, I have three PCs, one of which has WebODM, Node ODM and Cluster ODM installed. On the other two only NodeODM.
This part works now only to make sure the loads are connected through the work through the ClusterODM node to which the other two nodes are connected.
I’m trying to better understand the syntax to make sure to distribute the loads on the PCs that act as a node through the WebODM interface.
In practice on the nodes I put:

docker run --rm -p 3000:3000 opendronemap/nodeodm

Is this correct or do I have to specify something else?

I tried to change the parameters from the WebODM interface trying to figure out how to best distribute the workloads on the nodes, but I am still in the process of searching for a good solution.

I tried using:

max-concurrency for worker threads (do you mean the sum of the cores of the individual nodes?)

sm-cluster (by inserting the link to the node of the ClusterODM)

split (to divide the dataset into multiple workgroups)

On some datasets I get the low RAM memory error, but if I process them with only one node everything goes well.

How can I better distribute the loads (for example the RAM to dedicate to streamline the work in a shorter time without blocking the process?)