What's changed in LINCOMM 2.0

If you used the old LINCOMM servers, a few important things work differently in LINCOMM 2.0. This page explains the two changes most likely to trip you up: how you connect and how the servers share CPU and memory. It is meant to get returning users up to speed quickly during the switch. For step-by-step connection instructions, see the guides linked at the end.

LINCOMM is short for Linux Community Servers — a small group of shared Linux machines for running statistical software such as R, Stata, and Python on jobs that need more memory or time than a personal computer can give.

How connecting has changed

This is the biggest change, so read it carefully even if you skip the rest.

On legacy LINCOMM, you connected to a single address (lincomm.aae.wisc.edu) and the system sent you to whichever server was least busy. You could also land on more than one server at once.

In LINCOMM 2.0, you connect directly to a specific node. There are three nodes:

• lincomm-node01.aae.wisc.edu
• lincomm-node02.aae.wisc.edu
• lincomm-node03.aae.wisc.edu

The address https://lincomm.aae.wisc.edu is now a web page, not a server you log in to. Open it in a browser to see which node is least busy, then connect to that node by name.

Two more connection rules are new:

• One node at a time. You can be logged in to only one node at once. To reach work you left running, connect back to the same node you started it on.
• Your sign-in name now includes the domain. Where you once signed in with just your NetID, you now sign in as <netid>@AD.WISC.EDU (for example, jdoe@AD.WISC.EDU). "AD" stands for Active Directory, the campus account system. Your password is unchanged — it is still your NetID password.

How CPU and memory sharing has changed

Legacy LINCOMM did not limit how much CPU or memory any one person could use. One heavy job could slow the whole server for everyone.

LINCOMM 2.0 shares resources fairly between everyone signed in to a node. Each node now has 377 GB of memory, but your account gets a set share rather than the whole machine.

CPU is split by the number of active users

The available processing time on a node is divided among the people using it. If you are the only active user, you can use most of the node. If ten people are working, each share is smaller. This keeps any single job from starving the others. If your node is crowded, checking the https://lincomm.aae.wisc.edu page and moving to a quieter node can help — just remember you can only be on one node at a time.

Memory now has a soft and a hard limit

Your account has two memory limits, applied per user:

• Soft limit (44 GB): the level the system tries to hold you to. Going past it briefly is tolerated, but the system will start reclaiming memory and your job may slow down.
• Hard limit (48 GB): the absolute ceiling. A job that tries to go beyond it will be stopped.

If you have a genuine short-term need for more memory — a one-off large job, for example — you can request a temporary increase. Contact helpdesk@cals.wisc.edu with the details before you start the job.

Side-by-side summary