Closed cossatot closed 9 years ago
If you want me to schedule a meeting with Ron Young about running batches on the system just let me know. Looking at their wiki page is fairly intimidating, but I'm not scared to ask him questions. If you are playing around with and want me to ask specific questions, just let me know.
I don't think we need to set up a meeting. I have an idea about how the cluster system works now, I just need to see an example script. I though I saw one a while back on the ncsee wiki. I will look again.
Basically, they're just files that say 'run this sequence of commands, and do it for this username with this many nodes' or something. Pretty simple.
Sounds good. There is an example on the wiki. Let me know if I can help in any way!
You need to ask Ron where you should store results. I.e. you want a persistent folder somewhere, but it might not be on the head node. They probably have a special server just for that.
Also, see if you can get Ron to help you set up an example bsub script. It only needs to do one thing on the server (for example, 'bin/Pecube') but the important part, which I can't help with a lot, is to get the configuration right.
Sorry was out of town last weekend and didn't have a chance to answer these comments. I will send Ron an email, but first are we only running "bin/Pecube" or are we going to run "sh bin/run.sh" to create the VTK files? I know that we had discussed this a little, and I can't remember what you had said about minimizing the data output.
It doesn't matter, the command itself is arbitrary. We just need the configuration.
We need to create an example bsub script that runs Pecube. Then, I can make a Python script that modifies it for the new run parameters.