PSI-openstack people gave us the access to RHOS-01 which is nice because we now can have machines with SSDs to run our CI on. The catch is, we did not set it up perfectly fine, and we are going to need a fix for this at some point.
So far what we've done is to statically split our projects on the different CIs:
imagebuilder and osbuild are going to RHOS-D
composer goes to RHOS-01
But that's not good because, we can't keep the same quotas we were having on RHOS-D on RHOS-01. We can expect either having a 50-50 quotas split on the two platforms, or to stay only on RHOS-D (which we do not want) or to migrate everything on RHOS-01 with 70% of the quota we had on RHOS-D. I personally think that the first option is the better, but for that we need to have a smarter way of splitting the resources.
It would be better if would have the ability to split the resources inside a single project. Thus we would have the opportunity to run a job against RHOS-D while another runs against RHOS-01 in the same project. But to do that we need to make some modifications on the teraform script. Indeed teraform awaits some ENV variables that are containing the address and credentials of one openstack cluster not two. So we need to make it accept the two envs and make it able to load the proper credentials depending on where is the resource definition a job needs to execute its content.
PSI-openstack people gave us the access to RHOS-01 which is nice because we now can have machines with SSDs to run our CI on. The catch is, we did not set it up perfectly fine, and we are going to need a fix for this at some point.
So far what we've done is to statically split our projects on the different CIs:
But that's not good because, we can't keep the same quotas we were having on RHOS-D on RHOS-01. We can expect either having a 50-50 quotas split on the two platforms, or to stay only on RHOS-D (which we do not want) or to migrate everything on RHOS-01 with 70% of the quota we had on RHOS-D. I personally think that the first option is the better, but for that we need to have a smarter way of splitting the resources.
It would be better if would have the ability to split the resources inside a single project. Thus we would have the opportunity to run a job against RHOS-D while another runs against RHOS-01 in the same project. But to do that we need to make some modifications on the teraform script. Indeed teraform awaits some ENV variables that are containing the address and credentials of one openstack cluster not two. So we need to make it accept the two envs and make it able to load the proper credentials depending on where is the resource definition a job needs to execute its content.