So many changes happened in 2019. I thought I'd try to capture a few of them here.
In December of 2018 my long-running dumpy server started having problems. I "solved" this by getting an addressable power strip that has the capability of resetting itself. I configured it to ping the failing server, and when that failed a few times, reset that server's power port. It took a couple tests to get it right. I figured it would be a stop-gap, and I'd quickly move the services to the new server, the 128GB, 32-core, 3GHz beast It's been a year, and slowly, I've moved some things over there.
What really happened, though, is that a few months ago, the mainboard on the server finally went kaput. I threw up my hands, bought a new mainboard, CPU, and RAM. An hour later, and a bevy of updates, and the server was running again. It hasn't had an unexpected hiccup since! I kick myself a little because I first went cheap with 8GB of RAM, and the next week saw a sale for 16GB RAM, so I did upgrade that. Alas, the mainboard only has a pair of memory slots, so now I've got 8GB of RAM waiting for a computer to sit in...
Even with the server running well, I decided to continue moving sites to the new one. It has sat in the basement keeping time, and running one underused Jenkins server for a couple of years now. I finally got some time to rebuild my long-time hand-managed sites into Docker containers. I leveraged a Docker-aware reverse proxy that builds load-balanced routes to the separate containers. It's been an interesting experiment, and even with the smaller storage on the new server, it's not yet choked.
Now the system is generally static bits, and small apps running on the big (128G, 32-core) server. Databases and some file storage running on the slightly smaller (64GB, 64-core) server. Some websites and apps, e-mail, and centralized logging, are all running on the rebuilt (now 16GB, 8-core) server. Backups go to the HDD-heavy workstation (8GB, 6-core) A bevy of gigabit networking between them, connected to my 100mb/s Internet (promised gigabit there, too....still waiting). I'm eyeballing another new mainboard (with 4 memory slots) that I can stick my now extra 8GB sticks in, for a new workstation. I'll probably keep the existing HDD beast as a network backup.
I have spent a little time tightening my neglected e-mail server, too. Have amped up the anti-spam, including an external service for one domain that is hammered with UBE that makes it through my local filter (once I find a more aggressive configuration, and time, I'll probably adjust and drop the service; nothing against the service). I've tweaked a few relay and virtual domain settings, and added SPF and DKIM rules to all of them.
Upcoming plans are to move the remaining apps from the (rebuilt) old server to the new server. I need to finish configuring the load-balancer with SSL for the domains; I've got an SSL-enabled CDN, but its connection to the new server is not SSL. I've kept any sites or apps that really need SSL on the old server, where they are SSL to the CDN. But once the load balancer is working right with my SSL certs and dynamic Docker images, all of the static sites and local apps will be moved there.
That'll largely leave the logging and e-mail to the old server. Mostly the mail server does some UBE scanning to discard junk, and then forwards to other mail services (who also probably UBE scan, but when i don't do it first, I get hampered by UBE time-outs...). The mail server is down from its peak of thousands of users to just a dozen or so. Only three accounts use the server for mail storage. If I can offload that, then the server will be simply the log aggregator. Right now I'm using Splunk, but I'd like to break that down (and lose my decades of records) and replace it with an ELK stack instead. I prefer Splunk, but the Free Splunk has wide-open permissions and log input limits (that I rarely, but occasionally press). I've also considered either moving my Docker management to this server, or changing it up a little, and make this system also be part of my Docker cluster. Right now I just have the one container server.
For the workstation, there's only age reasons to update it. It's about as old as the old server was. it hasn't exhibited any issues, but its CPU is aged, and it's now the least responsive machine on my network. It isn't really a bother, as I tend to do most of my home computing on the servers now, and use my laptop more than the desktop, but the laptop is an aged MacBook Pro that doesn't support the latest MacOS; it'll soon fall out of favor. Same with my Mac Mini. Those three machines are about to hit some age limits, and I'll need to do something newer!