msdisme
commented
1 year ago Info from ticket: For our experiments (informed scheduling project), we need at least 4 nodes up to 16 nodes on 4 racks with cache nodes (I was not aware Peter needs 2 nodes). If we will not have nodes on the cache node racks, it does not make sense to keep cache nodes under HIL/BMI anymore. The whole point is to have the whole infrastructure available. For usage time, we are still in the development phase and it should take around one month to be able to run experiments.
I am not sure about the Cloudlab solution, since as the storage research group, we need constant access to cache nodes with NMVe drives (regardless of the project).
@pjd-nu what do you think?
Discussions here: Discussion in Ticket here: https://osticket.massopen.cloud/scp/tickets.php?id=1652 Slack discussion: Adding a link to the Slack discussion: https://massopencloud.slack.com/archives/GB7CT1NGK/p1684514142198949
Amin's D4N needs at least 4 nodes with NVMe case nodes and 16 client nodes
can we use neu-03-41, neu-05-41, neu-15-41, and neu-17-41?
what can we use for client nodes? Sharing neu nodes with Sahil.
if we can do it before July great if not we need to wait until end of September.