Geek dump, run away, run away!
If I was a normal person, much of the software and hardware issues I run into would just go away.
Instead, I invite these disasters on myself.
In a normal business or home, you would have computers for each of your workers. You would have some virtual machines on the cloud.
The work computers would be reasonably fast, have reasonable graphics, reasonable memory, and reasonable amounts of disk space.
The idea is to give your workers the right equipment to get out of their way.
My life isn’t like that. I have multiple virtual servers in multiple states from different providers. All of which need to be monitored.
At the office, my computer has good graphics, unreasonable amounts of memory, very good CPUs, and unreasonable amounts of disk space.
There are two more machines that are powerful enough to be considered “servers”.
Nobody really needs servers in the office space. I do things that make it reasonable to have those servers.
I run multiple virtual machines for testing purposes. Occasionally, it is just easier to toss up a new virtual machine than to try to run it on my standalone servers. I have a cluster of virtual machines running as a K8S cluster. There is a ceph cluster running to provide a distributed multipoint mounting system for those virtual machines.
So I installed Zabbix on a virtual machine. That virtual machine uses a ceph file system. This means that I can migrate that machine to any of my servers that has access to that ceph cluster. Which is very neat.
I had to learn how to write Zabbix templates to add monitoring of Amanda backup sets. Just got that working a week or so ago. And it has already paid off.
As I figured some of this stuff out, I added more and more of my servers and client servers to the monitoring load.
One of the things I added was disk hardware monitoring via ‘SMART’.
S.M.A.R.T. then told me that I had a drive that was running hot. Then it gave me a warning that a drive was failing.
Yesterday that drive failed. That drive failure affected my ceph cluster. Now, I found this out when I got alarms from Zabbix.
Meanwhile, I’m trying to fix my technical drawings. I was using the latest, greatest, version of FreeCAD. I was running it from a “snap”. Turns out that snaps run in a confinement. This means that I couldn’t get it to run any outside programs that were not approved. That was two+ days chasing my tail.
This caused me to revert to the stable release of FreeCAD. Which reset all of my settings. Which added hours to redoing parts of my blueprint.
FreeCAD can record a frame for each step change in a variable. That’s cool. I’ve made a couple of short animations using it. But what I really want to do is to have it send each step to a ray tracer. Which I’ve not figured out how to do yet.
So it is now 0115, I have almost recovered from the bad drive issue. I’m just waiting a bit longer to be able to reboot that server and take the drive out. I can’t do that yet because ceph wants that server to be up for a bit longer.
If I get very very lucky, I’ll be done with this shortly.
Comments
4 responses to “When things work?”
This is kind of tangential, but do you know of any simple intuitive introductions to FreeCAD? I’m trying to come up to speed on it, and it’s just not intuitive for me. I am a long term (over 20 years) user of Blender for my work, but my habits from Blender just don’t transfer to FreeCAD at all. I recently retired and have started learning machining for some home projects, just to be able to do it (my great grandfather was a blacksmith, my grandfather was a machinist, and my father was an engineer and machinist). Blender is great for demonstrative and creative work, but it’s hard to get the measurements right for machining. CAD and CGI seem to be different skills.
Indeed they are different, as I found out when I tried to go in the opposite direction.
If you know what you’re trying to create, but the FreeCAD UI isn’t helping, a possible solution is to use the scripting machinery. About my only use of FreeCAD so far is a complex CAD model of a spacecraft (Rolf Nelson’s “Tajemnica” — see “The stars came back”). What worked well for me is to treat FreeCAD as a CSG tool, and use a Python script to describe the elements I want and place them in the model.
For extra fun I also included code to generate the PovRAY version from that, so I can do ray-tracing pictures. The Python scripting is reasonably well documented in the online documentation.
Sorry to hear, hope you get it cleared up.
.
“Nobody really needs servers in the office space.”
.
That depends on the office. Engineers and draftsmen often need pretty powerful systems for their CAD work, depending on how complex the final assemblies are (parts count, number of features, etc.) So too do many scientific users, although the emphasis tends to be a little different. Both use graphics coprocessors (the latter for the raw computing speed more than, well, graphics), and far more memory than the average office user typically needs.
.
Some of our people actually have – and need – mini-clusters in their offices.
.
It’s an ongoing argument at work … Time sunk in getting IT to approve task-appropriate computing resources, or buying a standard “workstation” that is about 80-90% of the way there. The science users also typically need to write their own software, which triggers its own brand of conflict with IT security.
I was a Trash-80 adoptee in the 70s with a cassette tape for data storage (why isn’t that damned asterisk blinking?!!!); I learned BASIC from a monthly magazine; I upgraded my computer to a whopping 4K, yes, K, with soldering iron at the keyboard (where the CPU was); my sons grew up hearing modem connections as I hosted a bulletin board, long before the Internet; I inked my own cartridges for years; I’ve never left computing — just the wife and I and there are several machines in the house.
But I have told my sons that electronics left me in the dust a couple of decades ago.
And reading posts like this convey that I am not in the dust. No, I am in the mud!! 🙂