jr conlin's ink stained banana

A while ago, i had a discussion with a coworker about their boss. My coworker was venting a bit about a number of patterns that i recognized from my experience with various managers.

First off, let me say that i’m not a manager. i was a manager for a while and was terrible at it, so not being one to want to repeat mistakes, i haven’t pursued being one since. Being a good manager is incredibly difficult and requires a number of talents that are not included in WSJ Productive Resource Engagement books or whatever the current trend is.

What i do have a lot of experience in is managing my manager. Each manager has their own quirks, but there are certain types of managers that i’ve learned how to train and deal with.

Let me talk a bit about Process Managers.

Process Managers are folk that are in love with the process of management. These are folks smitten by things like JIRA, AGILE, SCRUM, and all the other things that either add or remove things from various checklists that they fetish over. These are people who “want to see progress” and expect their reports to have complete, analytical breakdowns of tasks done within minutes of being assigned. These folk do not do well with things like “discussion” or “revision”, since that requires them to alter the carefully crafted charts and diagrams that they spent all week putting together to impress their own manager.

Some folk might feel that these sorts of managers like to micro-manage, and a number do, but mostly, they’re looking for the dopamine hit of “number goes up” (or down, but basically not “stayed the same” so that they can show “the process works!”).

As a software engineer, we know full well that Life Does Not Work That Way. Asking how long a given task might take is like asking “how long will it take to carve this statue?” You have a rough idea, and can show progress, but as an artist, you have no idea what’s actually in the stone and if you it something unexpected, you have to deal with that. Sometimes you need to alter the design a bit. Sometimes you have to start over. It’s the nature of the art, and that is completely incomprehensible to the Process Manager.

So, how do you manage a Process Manager? You feed them the dopamine they’re looking for. Give them short updates where you specify the work that’s been done and the very short scale work that lies ahead. Give them little things that they can check off of a list. If you need to alter a design or the scale of something changed, give them a new checklist of things to tick off. Track Everything via the ticketing system of choice (Jira, bugzilla, Postit notes on the cube wall, whatever). They’ll grouse, but you gave them delicious mental candy, so they won’t stay mad for long.

Play the game they want to play and they stay happy.

On the other side, realize that they are not invested in your growth and career unless it maps to some form of progress they can chart. So, if you want to go to a conference or learn about something, you need to absolutely qualify it with all the bells and whistles they want to see. Put together an Impact Assessment that highlights what percentage points of improvement you expect to deliver or whatever.

(bit of a side note: Process managers tend to think of reports as ‘resources’. You know, like the copy machine or coffee maker. Mind you, when the coffee maker has a bad day, i replace it, because coffee makers very seldomly self correct. A coffee machine doesn’t perform poorly because it’s kid was up until 3 AM, or because a parent just got a cancer diagnosis. People do, and they’re not going to tell their manager or coworker about that for lots of reasons. People have off days and weeks, and a good manager knows and works with that. The coffee machine also doesn’t angle to get the copy machine’s job, but that’s a different matter.)

i kinda spelled all that out for my co-worker, with the added point that “all of this is temporary”, none of it is actually personal, particularly since the senior engineer on the project has zero complaints, and frankly the stuff the Process Manager is complaining about is stuff like “too much discussion in PRs”. Uhm, you want discussion in PRs. That’s documentation. It also means that folk aren’t just rubber-stamping future bugs into the code. i’d rather there be good conversation in a PR than a bunch of issues being re-opened.

i also gave a bunch of other tips for how to deal with a Process Manager, and maybe i’ll write those up sometime.

A key aspect of a good engineer’s life is dealing with non-technical issues like these. A key aspect of a senior engineer’s life is giving junior folk the insights needed to become a good engineer. Dealing with iffy managers definitely falls into that category.

Note: The following is sweary.

Let me first point out that all of this is mostly self inflicted.

Let me secondly point out that i think i understand the appeal of laptops now…

The Triggering Event
i had an Acer Spin 13 for years. The idea being that i have a cheap-ish “walk around” for when i travel, or i’m sitting on the couch. It ran ChromeOS, mostly because i could install a Linux VM on it and run apps like VS Code and Firefox that way. The real beef was my custom built tower: 64GB of memory, way too much disk space, and an i7. It was my everyday workhorse. i mean, sure, that i7 was maybe gen2 or so, and the MB was subject to Heartbleed and other attacks, but it worked well enough.

So, for various reasons, i decided to get myself a Framework PC.

This PC is amazing. It’s exactly how PCs should be made and offered. Yeah, dropping a K on a “walk around” makes it a wee bit more concerning should i lose or damage it, but the form factor, weight, power, and features definitely make it worth it. Installing Lubuntu on it was a dream.

Then, just for giggles, i installed Rust on it and compiled `autopush-rs`.

Oh, dear.

What normally took minutes on my “powerhouse”, took just shy of a minute on my laptop. It was time to upgrade the beast.

Rolling the dice
Doing a modern build is like trying to put together a well-rounded stereo system. It’s possible, but there are so many scams and “BUILT FOR X-TREME GAMERZ” bullshit that it requires a ludicrous amount of attention. As noted elsewhere, i’m not a fan of a 3.2GHz clown cars.

So i put together a system that, i felt would work ok:

• a 12th Gen i9 Intel processor (Not the latest and greatest, but that’s ok.)
• a z690 Motherboard The last motherboard i got was an MSI, and it was both LED free and worked nicely.
• 64GB of DDR5 memory
• i even spurged and got some extra thermal paste, because i would be reusing the old cooler.

i planned to reuse as much of my old system as possible (ha-ha).
i mean, i figured the worst part would be (Ha Ha) having to re-authorize Windows 10 (HA HA).
How bad could it be? (HAHAHAHA sob sob)

i pulled my car out of the garage, and set up a computer surgery center. i even told my wife, “i should be done in around 4 hours” (Oh, sweet innocence…)

The Struggle, Part 1

After cleaning out the case and wiping out the fans (which i do every year anyway. Leaf blowers are amazing for that.), i got to work carefully extracting the old motherboard and cooling stack from the old system. It was annoying, of course, but went fairly well. i wiped the old thermal paste off of the bottom of the cooler and the CPU with a bit of rubbing alcohol, and set them to dry.

Then, it was a simple matter of setting the CPU (why the hell is there so much packaging?) and memory (wait, the slots are 1 and 2 not 0 and 1, and which ones are the primary?). Now for the cooler. Wait, where are the mounting pots? There are holes, but… Oh for fuck’s sake.

So, apparently modern motherboards no longer have pre-built mount points for coolers. Awesome.
So, off i go to the local computer parts place to get a new cooler. Oh, hey, the old power supply was kinda noisy (why don’t they offer db ratings for those?) let me swap it out for a newer quieter one. Power connectors are standardized. (Ha-ha-ha)

The Struggle, Part 2
So around $300 later (cooler and power supply, i set about building things again. i fix the base, apply some paste, and get to mounting the cooling– crap, this thing isn’t balanced and keeps slipping off the– dammit, i have to get the grodding mounting screw through the fan ducting– crap it fell out? Where the frak did that screw– DAMMIT the cooler slipped off again!? Wait, what was that snap?

Turns out that one of my memory chips was about 1mm too tall. Yes, apparently, that’s a thing. i mean the bit that cracked was part of the decorative X-TREME GAMERZ ONLY crap, so nothing critical, but still, might want to get vertical dimensions for your DIMMS along with base clearance for your CPU coolers from now on.

Well, no big i’ll just run a cable from this to the garage monitor… that’s DVI, because it’s old. Fuuuuuu….

i brought the system into my office to fire up because that’s the closest HDMI/DisplayPort monitor i have.

Hey! It boo– WARNING! MEMORY IS NOT IN OPTIMAL SLOTS!

Well, taint truffles, it’d be nice if you had something that actually NOTED THAT ON THE GODDAM BOARD or HAD A FECKING MANUAL THAT INDICATED THAT now wouldn’t it?

Insert another hour (yeah, hour) of me removing the cooling stack, moving the memory to the proper slots, and reattaching the stack. By this point, the thermal paste is a complete mess and i’m fully aware that i’m going to have to reapply it eventually. THAT WILL BE ANOTHER DAY.

Fine, it’s all bolted up. Finally, i can power it up and see if the License key still works…

The Struggle, Parts W & L

Well… shit. That’s unexpected.

Apparently, the new config is causing whatever bits of clever Microsoft did to get Hyper-V running to fail dramatically. i can’t boot it at all, even in safe mode. i go through the various recovery mechanisms, and no soap.

Ah well, i was kinda prepared for this, so let me grab my LiveDisk copy of Lubuntu 22.04 boot, and then install to the partition, restart and…

Wait, what? You have 64 GB of memory. how the hell can you be out of memory just booting up!? You’re Fscking Linux, for Crissakes! You run on crap that has 64MB of memory.

insert 80’s montage of me scrubbing through various docs and sites that basically work out to :man-shrugging:, and me adding increasingly obscure swear words to my growing list There’s some weird problem with MSI motherboards, built in video, and overly large boot images being generated.

Wait, what? Turning off `silent` from the boot options fixed it!? SERIOUSLY!?, Well, pisskidney, sure, i’ll let it spew it’s brains out on boot.

The Struggle, Part ?
So, as of now, i have Lubuntu on my main system, and it’s compiling autopush-rs in about 30s, which is a literal magnitude better than it was prior. i have Windows currently updating from an original install going in a VMWare instance, where i’ll confine things like some games and other windows only sort of things.

My main system is currently running Lubuntu, and i’m basically building it up as my main, which means installing lots of stuff and moving bits over. i still plan on swapping the guts out of my garage server for my old motherboard, power, and cooling fan, but that’s going to be some other weekend’s project.

But yeah, definitely starting to see the appeal of laptop only lifestyles.

A while ago, someone pointed out that you can get an induction burner from IKEA for around $70 i’ve been induction curious, since i’ve never tried one before, so i bought one.

Wow.

So, with all the discussion about induction and people coming for your stoves, no one bothered to explain to me the mechanics. They’re… not what you think. i’ve been cooking with a gas stove for decades, and this is very different.

Allow me to try and best explain what’s going on.

Let’s say you have a resistive electric cooktop. Electricity goes in, then heats up the coil using resistance to the point where it generates radiative heat. That heat, depending on the make and style, then either gets passed up into the cooking surface or contact heats the pot. Since it’s radiative, heat goes in all directions so only a percentage of the heat goes into the cooking vessel.

Gas works pretty much the same way, only more so. It combusts producing heat that is then dumped into whatever happens to be near by. Since convection is a thing, most goes up into the vessel.

It’s important to note that the heat in both of these cases can go around the vessel. Gas is far worse at this than electric, and anyone who’s cooked with gas will tell you to mind the flame so that you’re not making the pot handles too hot.

Induction doesn’t do any of that. Induction produces heat in the vessel directly via fun electromagnetic forces. This means that the wattage you ask for, goes into the vessel with very little being wasted. That’s absolutely not the case for resistive electric and gas.

When i boiled water using a pot i’ve used hundreds of times in the past, the handles were cool. Possibly for the very first time in that pot’s lifetime. As a fun bonus, the surface of the cooker cooled rapidly once the pot was removed because it was not the source of heat.

i’m interested in trying it with a few other cooking styles, but i’m going to guess that it’ll do just fine.

i am now very sad that induction cooktops were crap when i redid my kitchen 10+ years ago. Joe can absolutely come for my gas stove.

Much like can openers, things have absolutely improved.

Maybe it’s the end of the year and hearing one too many bad versions of “Auld Lang Syne”. Maybe it’s the fact that i’m on day 6 of COVID. But for some reason i was thinking about Mr. Scott on the Enterprise from Star Trek.

i started thinking about what a massively complex system a 400 person capable interstellar space ship, complete with armaments and transportation systems, fueled by carefully exploding matter and antimatter, which produces pure energy at an unfathomable scale must be. And yet, you’ve got basically one guy who somehow knows and can repair all of it.

i mean to micro-geek me sitting on my bed trying to hold the UHF antenna just right on the black and white so i can watch reruns after-school, sure. And for a bunch of writers on a fairly low-budget TV show focused on telling stories about societal ills dressed up as “Wagon-train in the Stars”, you bet! But for grown up me who’d be about the right age myself? Oh, hell no.

One of the hallmarks of being a Senior Engineer anywhere is having a firm and complete understanding of what you do not know. It should spread out before you like the vast, endless plain that it is. You might recognize features like distant mountains or far off towers, but that’s probably about it. You should also know who to ask or rely on to navigate those particular bits when you need to, and also recognize that unless you’re working with that stuff frequently, you’re not going to instantly become a domain expert.

So, yeah, having one guy who basically is responsible for keeping a multi trillion dollar machine flight and fight ready at all times, as well as will be on call to unplug the Command Head after the replicators finish Taco Tuesday? Yeah, nope.

For work, i do what’s become called “DevOps”. My company defines that as building scale-ready back end services that can be deployed and maintained for extended periods. This differs from our SRE (Systems Reliability Engineers) who’s job it is to make sure that the systems we built are deployed and operational. i’ve sometimes said that they’re the “Check Engine” lights of our org. They’re critical players that monitor and understand our running systems and how they relate far better than anyone else. i could not do their jobs and i’m always exceptionally nice to them. For anything that’s truly broken, their job is to file a ticket or call me and make it my problem to fix.

Same with our Test Engineering folk. Their job is to make sure that the things we build and operate run consistently, validly, and efficiently. They are our highly skilled, trained chaos monkeys who actually help fill out reports. They look to make sure that we’re not just doing “happy path” testing, but introduce appropriate noise and hostile behaviors that we miss because we just want this crap to work (dag-nabbit).

Each group knows their skills and shortcoming. Each group happily works with the others to get a fantastic final result. i would no more dismiss these groups as “useless” than i would dismiss a surgeon because “My G.P. knows how to medicine. How hard could it be?”

i suppose this also gets into the whole myth of the 10x engineer, (Note: a 10x engineer is someone who does a job that is only .1% defined.) but that’s a different discussion.

All i can think is that if Mr. Scott has to do some sort of repair or fix on something, someone else is dead or incapacitated and that a third person may be needed to spend a day or two repairing the “fix”. i mean, yeah, i can deploy a system on a set of servers that could run for some period, and have in the past, but i’m not dumb enough to think that it could handle several times the population of the United States without me personally funding the next trip Jeff takes to the stratosphere. i’m sure that there are whole catalogs of “Scotty Fixes” pictures at Star Bases around the Federation showcasing plasma conduits wrapped in chicken wire and silicon tape.

Mr. Scott was my favorite character on Star Trek for a slew of good reasons, but the further i get in my career, the more i appreciate the unseen folk who made it their jobs to make sure that Mr. Scott barely knew the systems they worked on.

So, stemming from my departure from Twitter, i decided to go for the third phase of being on Mastodon, and set up my own server. Mastodon is interesting because it’s pretty much like email. This may become a problem in the future, but i also tend to post a lot and have a fair bit of media in my posts, so paying for a server instead of just dropping a few coins in someone else’s bucket seemed like a fairer option.

(Even though i said i would never set up a server because “who wants to deal with that nightmare”?)

## The failed attempt

The first try was… not successful. i still don’t quite know why. i got an instance, and tried doing a docker-compose style install. Everything mostly went correctly, except that i could never properly federate. Requests would go out, and remote hosts would sometimes get them, but no approval responses would return. There were many rabbit holes to be fallen into, and i’ve seem to have found most of them.

Realizing that my life should be more than debugging this one, particular instance, i decided to try again on a different service.

## The successful attempt.

DigitalOcean offers a “1-click” setup for Mastodon. It’s more than one click.

i won’t go into to the same detail as many of the step-by-step guides, but here are some bits of guidance for the things not covered in the tutorials. (These are also here for my benefit, if i need to do this again.)

• i probably over provisioned. My “free”-ish failed attempt was going to cost me about $20 a month to run. Using that as a guideline, i picked a config that has 2vCPUs, 2GB of memory and 60GB of drive space, which should run me about $18 a month. (i’ll update this later to show actual cost.) Rather nicely, DigitalOcean appears to automatically update packages on initial install, but looks like i’ll need to rig up an automatic package updater, or just add it to my list of machines to manually update.
• Since my image has 60GB of storage, i didn’t get any additional block storage. (Block storage is added as a disk to the image, so you’d need to do some work to wire it into your mastodon configuration.)
• i also made sure to load up my dev public SSH key so that no root password login was possible.
• i followed the getting started steps for the app. i was a bit surprised about you SSH in as “root”, but i’m sure they have their reasons.
• Once i verified that things are working, i copied the ~mastodon/.env.production file to my local machine for archiving purposes.
• After that, i set about updating the droplet’s Mastodon 3 version to Mastodon 4 using this very well written guide.
• That done and proofed, i set about looking at pre-banning the nazis. This page lists a number of bad actors as well as some moderation automation tooling. i’m currently looking at integrating the Hackyderm Admin Blocklist tool, but i just sorta cheated and added the initial batch by hand.

Hopefully things work. FWIW, i don’t really have any plans on letting folk join my instance, mostly because of the HUGE increase in work being an admin entails. Right now, it’s a toy for me to amuse myself with, and i might go back to using a proper server in the future (that’s why i generally maintain older accounts).