SysAdmin Weekly #18: Craft, Shovelware, and the Cost of "Good Enough"
Why modern software keeps failing operators and why SysAdmins pay the price
⏩ TL;DR – This Week in SysAdmin Land
· The newsletter is evolving! This means fewer link dumps, more opinion, commentary, and real-world SysAdmin pain from the trenches.
· Theme of the week: Craft vs. Shovelware and why modern software feels more brittle, less thoughtful, and far more hostile to operators.
· A real-world rant with receipts: Microsoft Authenticator quietly falling over once you manage a lot of accounts… and the only fix being “nuke from orbit and start over.”
· The latest SysAdmin Weekly Podcast digs deeper into “good enough” software and why SysAdmins are the ones paying the price for it.
· Community Signal highlights thoughtful takes on whether software is actually getting worse. Plus practical Azure Local update guidance you’ll actually need.
· Other links worth your time this week include one rock-solid security podcast and one non-technical sanity-preserver (because balance matters).
· Looking ahead: plans to spin up GitHub Discussions as a low-friction, SysAdmin-friendly place to keep conversations going beyond the podcast and newsletter.
An Update on the Newsletter
First off, I’m sensitive to the fact that we’ve been away on this newsletter for quite sometime (4 weeks?) and I’m also very aware that I’ve provided MANY “Updates on the Newsletter” like this over the last month or two. The fact of the matter comes down to two things:
1. With the day job, the podcast, and a VERY time intensive side-project I’ve been working on, getting to this newsletter has been inconsistent and difficult.
2. I haven’t been 100% feeling the “vibe” of the newsletter format we’ve been using up until this point. Statistically “Link-dump” newsletter are one of the most common and while, yes, they can provide some value depending on the quality of the links, dumping a collection of links each an every week just to provide a collection of links doesn’t produce CONSISTENT value that I personally find worthy of your time.
Moving forward this newsletter will serve as much more of a commentary and opinion-based newsletter remaining laser-focused on SysAdmins and the challenges they face in the trenches every day. Will there still be some links? Sure, but the format will change to content that is much more conversational in style which provides a better backdrop for me to produce that CONSISTENT value that I want to provide this community.
If you have comments or concerns about the new direction, do let me know in the discussion section below!
From the Console
Our theme for this week’s edition of the newsletter focuses on “Craft vs. Shovelware”. I’ve been in the industry long enough now to recognize that software seems to be more brittle and problematic now than it used to. The amount of times that I’ve had to employ some sort of workaround to a seemingly simple-to-solve bug in order to keep production live has certainly increased over the years. And yes, our environments are wildly more complex than they used to be what with cloud and hybrid deployments and yes, that introduces EXTRA chances for breakage. This does not however account for cases where applications operate in a way that is woefully unhelpful to the user.
I’ll provide my own example on this one. I have been a LONG time user of the Microsoft Authenticator app for MFA and Identity services across a number of tenants in Entra ID and various 3rd party accounts. As of this writing I have around 43 different accounts managed within the Microsoft authenticator app. One day, I tried to add account number 44 to the app and I get hit with a big nasty “device token error”. Long story short, it seems I’ve hit a community-experienced, but non-Microsoft documented issue with the authenticator app where either there is a some soft-limit on number of accounts or there is underlying token corruption. Either way, I am unable to add any more accounts until I rip and restore ALL 43 accounts currently in my authenticator app….. needless to say… I was NOT impressed.
As mentioned, Microsoft has never documented a maximum number of accounts supported by the Microsoft Authenticator app (that I’ve been able to find). However, there is substantial real-world community evidence that the app does not scale reliably for users managing dozens of tenants / accounts. Admins frequently report that once they reach a high number of accounts, Authenticator begins failing to add new accounts or register push notifications, often producing vague registration failures or “device token errors”. One commonly cited message states: “Error changing device token. You must delete and reactivate your accounts to correct the issue,” effectively forcing a full MFA re-registration across all accounts.
Microsoft’s public documentation focuses instead on per-user Entra ID limits (such as the five registered authenticator devices limit), while offering little guidance on app-level scalability or token exhaustion. Community discussions on Microsoft Q&A and Reddit consistently describe these token-related failures and the same recommended remediation of “delete and re-register” suggesting an undocumented Authenticator scalability or push token management issue rather than isolated user misconfiguration.
Which brings me to my point of this whole rant about software quality in the modern day and user-expected app behavior. I’m completely fine with imposed limitations on software (such as max number of supported accounts) as that is something that I, as a SysAdmin, can plan and architect for. What I would have expected in this situation, if my suspicions about scalability and token issues are correct, is a warning at ~30 accounts “Max number of accounts reached”. Instead we get silent breakage, and to quote Ultron from Avengers: Age of Ultron,…“Boom… the end… start again”.
All this is to say, software development needs to once again be given the time and consideration to be a “Craft” again. What we currently seem to have is a hastily built software “content-pipeline” that is designed to push features extra-fast and squeeze maximum revenue from every byte. Something’s got to give.
And now, back to our regularly scheduled programming.
🎙 The Latest on the SysAdmin Weekly Podcast
This week I had regular co-host Eric Siron back on the show to discuss the very issue of “Good Enough” software mentioned above. You’ll find that it’s deeply attuned to this week’s theme.
SysAdmin Weekly Episode 034 Thumbnail Image
Episode: “Good Enough” Software is Ruining IT (and SysAdmins are Paying the Price)
Topic: Software enshitification has been a real issue in the industry lately and there are real consequences (both spoken and not) of that fact. One of those consequences is the fact that SysAdmins are usually the ones that are facing the full brunt of poorly written software that was deemed “good enough” at some point. In this episode Eric and I (Andy) discuss the state of software in our industry while lamenting a time when software was a craft, and NOT just a content pipeline.
Episode links below:
· SysAdmin Weekly on Apple Podcasts
Community Signal
While researching the aforementioned issues regarding this week’s theme. I did happen upon some useful community posts along the same vein or similar:
This older post on the Stackoverflow Blog by Issac Lyman that asks the question “Is software getting worse?” Issac has a very fair and reasoned look at the question, and even though I’m not a developer, I greatly enjoyed reading it. I highly suggest you check it out.
I also wanted to include some practical SysAdmin help from the community this week, starting with How to Keep your Azure Local Instances Up to Date by Thomas Maurer! More SysAdmin’s are finding themselves in a situation where they’re starting to rollout Azure Local, so this knowledge is a MUST have.
Other Links Worth Your Time
As I’m a regular podcaster, I often get asked which podcasts do I regularly listen too. I’ll provide one of my favorite technical podcasts, and one more…. hobby related.
My technical podcast recommendation is very security related, and I’ve mentioned it on SysAdmin Weekly several times. That is the Security Now Podcast featuring Steve Gibson. Steve does an AMAZING job of explaining cybersecurity topics and when you listen to an episode of this show, you’ll learn more about the given topic then you ever thought possible! Highly recommended!
My NON-technical choice has to do with my long-standing gaming addiction. I’ve been playing video games since I was old enough to hold a controller and I love listening to Remember the Game. Adam (the host) does an amazing job of keeping things interesting and funny. If you’re a long time gamer like me, you won’t regret putting this one on your list.
Building the SysAdmin Weekly Community
A final closing note on my efforts to continue developing our growing community. It’s come to our attention that we need a place that is quicker when it comes to communications. Something that can be used for announcements as well ongoing discussion. While we could certainly comment on each episode posting as well as the newsletter comment sections here, the goal is to have a centralized location where most of the community conversation happens.
As it stands, I’m currently looking at using Github Discussions for this purpose. Most SysAdmins will already have a Github account, and it doesn’t require a new app to be installed, so the barrier to entry is low. Be on the lookout for this new hangout in the coming weeks.
Retro SysAdmin Fact of the Week
Don’t forget. For years, many enterprise systems treated a two-digit year as “good enough” and it could have taken down the world.
The Y2K problem wasn’t a mysterious bug or media hysteria; it was the logical outcome of early storage constraints and yes, short-term thinking. To be fair, saving two bytes per record mattered when memory and disk were measured in kilobytes. Meaning developers stored years as YY instead of YYYY. This “temporary workaround” quietly became the norm. When that calendar rolled to 2000, systems didn’t fail because they were old, they failed because, let’s face it, no one expected them to still be mission-critical 10, 20 or 30 years later. Y2K remains one of the most successful IT and SysAdmin remediation efforts in history.
See? The concept of “we’ll fix it later” has been around for a VERY long time.
Until Next Week
Thanks for reading this week’s edition! If you haven’t yet, be sure to hit that subscribe button. It’s free and you’ll get useful, maximum-uptime quality content in your mailbox every week to help you in the SysAdmin trenches. Also feel free to share with another SysAdmin. For real…. we want to help as many people out of unintended systems outages as possible.
Additionally you can always find all the links associated with this newsletter and the podcast at www.sysadminweekly.com.
Until next week! Stay Frosty!
-Andy
SysAdmin Weekly