February Newsletter: Legacy Retirement Imminent (due March 31)

Hi, everyone!

Hope you’re enjoying your February so far!  It’s almost over, but I’m actually getting a newsletter out in the month I intend it to come out, so I want to acknowledge that and pat myself on the back a little bit.  Good job, Craig!  Thanks, Craig!

Okay, on to the important stuff.  I want to focus this communication on the retirement of the Legacy MCS computing environment.  This is the Unix computing environment that’s been running in MCS since before I joined Argonne over 25 years ago.  We’ve missed so many previously announced deadlines on turning this stuff off, and now it’s reached the point where the hopes, prayers, and band-aids that were keeping it running are starting to fade, go unanswered, and peel off in an unsightly way that makes you want to rip them off and be done with the whole thing. The system that manages the accounts in Legacy has been orphaned for some time, and we’ve lost the ability to allow people to set their own passwords, upload SSH keys, and other trivial functions for some time.

Earlier this week, we stood up a new login server for Legacy called “homes-legacy.mcs.anl.gov” and redirected the old “login.mcs.anl.gov” (aka login1.mcs.anl.gov through login4.mcs.anl.gov) to it.  This host is not accessible outside Argonne, much like the older legacy login nodes it replaced.  In order to login to it, you still need to be on the VPN or have a GCE account.  If that’s a host you’re still using, this message is very prominently targeted at you.

We’re still in the process of migrating some other projects that rely on Legacy authentication and storage to GCE (such as CANDLE, SEED, and older AI/ML systems), but we now need to focus on turning off general access to the legacy file server.  This is the server that serves up the home directories in Legacy.

Please see the migration instructions for migrating from Legacy to GCE. gitlab and xgitlab are now internal-only, and if you haven’t migrated to the new git.cels.anl.gov, there are instructions there for that.  The bottom three (home, project, and web directories) are the important ones to pay attention to.  Please take a look at the instructions, and try to get any important data migrated to GCE.  At the end of March, access to Legacy filesystems will generally go away.

If you have a particularly large home directory, we can help you determine what should go into project space vs. home space.  As noted in our documentation, GCE is using optimized solid state storage for home directories for improved interactivity.  Storing terabytes of data in the homedirs would get very expensive very quickly, so we strongly encourage storing anything project-oriented in a project directory.

If you need help with any of this, please reach out to us at [email protected] and we’ll do what we can to assist in making sure things get migrated smoothly.

After the legacy storage is retired, we’ll do a last backup of the data to tape.  Then after 90 days we’ll repurpose the legacy fileserver into GCE to provide more storage there.

Starting the week of March 20, I and other Systems team members will start hosting regular “office hours” online, where anyone can drop in with a question or watch a demo of how to do various things in GCE.  I’ll post the schedule in Slack and here on the blog in early March, but if you have topics you’d like to see covered, please let me know and we’ll prepare something for it.  To start, we’ll experiment with huddles in the #cels-systems-helpdesk channel, but we can break it out into a different channel, or even Zoom or Teams sessions depending on how well it goes.

We’ll cover topics like effective use of SSH, streamlining access to resources, effective use of project vs. home directories, and other topics to help make using GCE as effective as we can.  I expect these to be interactive, and the feedback we get during the sessions (or after the fact) will allow us to adapt things accordingly.

EDR/CrowdStrike Update: As a brief followup on the last announcement for CrowdStrike rollout, we’re making progress.  Managed and co-managed machines will get the updates automatically with no action required on your part during March.  We’ll send an announcement prior to it going live and let you know what you can expect. I’ll also be sending out instructions on how to apply the required pieces to any completely self-managed machines next month once we get it down to a simple download-and-install script for the various platforms.

That covers the important topics for this month.  As always, please feel free to reach out to us at [email protected], on slack at #cels-systems-helpdesk, or via phone at 630-252-6813.  There are links to reach us via our documentation as well.