How do you answer questions for which an answer doesn't technically exist

Posted by Educational_Sink_535@reddit | sysadmin | View on Reddit | 25 comments

Every now and then, random issues come for which there are no answers -- at least no answer within the scope of what I am able to check at the moment.

One case, someone came up to me that they need access to Teams. Access was granted and they attempted logging in. After getting through M365 SSO, they got stuck on the Teams login screen. Literally no button worked. After beating around the bush -- Incognito window, the works -- had to tell them to just try again later in like an hour. Lo and behold, it worked 😭

Another case, a Jira automation just didn't trigger for no reason, while it had been triggering fine for several previous runs. I just re-triggered it and boom, it worked. Same experience with an Okta workflow.

And then there are user-centric cases. A user walks up to me one day and swears that their password they've been using for months is just not working. Upon checking the logs, the last successful login was 3 days ago. Over the weekend, they didn't attempt logging in and the day they come to me (3 days later), the logs shows their incorrect password attempts. Clearly, from the logs, there is no activity that shows a password reset took place. If the user swears they are trying the right password they've "used all their life", how do you even begin to argue such a case without looking stupid for having no explanation. Do you blame the system, the user, or a glitchy universe?!

In cases like these, users / bosses expect you to have an explanation as to what is going on, to point the finger somewhere. But how do you convey that you have no explanation without sounding like you are just winging your job / clueless? I mean, in a case like this, what am I going to say, that it's really just a glitch / gremlin? And leave it at that? If I try to blame the tech/system, by ruling this as a glitch, I can't help but always have that feeling of inadequacy, that feeling of guilty that: "I admin this SaaS, I should know it like clockwork and I have no explanation for what just happened".

[-]

MediumFIRE@reddit

One of the most freeing things to say is "I don't know". I find people trust you MORE versus giving an answer they kinda feel in their bones is b.s. That doesn't mean I still don't try my best to find an answer.

If the user swears they are trying the right password they've "used all their life", how do you even begin to argue such a case without looking stupid for having no explanation. Do you blame the system, the user, or a glitchy universe?!

I've had some variation of this over the years. One time someone said this to me and I got to eye level with the keyboard and watched them type the password and without realizing it, they were pushing the space bar each time before enter.

[-]

2_Spicy_2_Impeach@reddit

Yes. Saying you don’t know builds trust. It’s just if you say you don’t know then don’t investigate is where it can be bad.

Sometimes stuff happens. As long as it’s not a reoccurring issue causing an actual issue, you’re good. If they don’t like your answer due to a vendor issue, we can always hop on a call and hear it together again.

Eventually leadership should trust you. And if not, look elsewhere. I left a job due to that. Suddenly my opinions are valid because of the domain my email address has?

[-]

Ssakaa@reddit

I've had some variation of this over the years. One time someone said this to me and I got to eye level with the keyboard and watched them type the password and without realizing it, they were pushing the space bar each time before enter.

That's a real mess... when they just got a new chair, and suddenly aren't pushing space before enter each time now...

old_cypherpunk@reddit

Explain how RAM works on a molecular level. They'll understand that it's a miracle that anything works on a computer.

himji@reddit

DNS

childishDemocrat@reddit

I don't know but I will find out.

luke1lea@reddit

And then leave and never talk about it again cause no one really cared all that much to begin with

RAMSxAI@reddit

Always a reason why, not always a reason worth chasing.

Anymore it is either

1) Microsoft 2) SaaS Vendor 3) Caching between the two

As for the password, not worth trying to prove them wrong, just tell them you need to reset the password and hopefully direct them to a self service link.

Even when they ask why something happens, only select few care to listen to the answer.

Sweet_Mother_Russia@reddit

Have you only been in IT for like less than a year?

Users don’t give a fuck about why something works. They just want it to work. They don’t need an answer.

“Looks like there was a little glitch there and it seems to have resolved. Let me know if you have any more problems in the future!” And then just pray it doesn’t happen again basically.

For the idiot who locked their account typing their Password wrong? Tell them that there appeared to be a communication issue on the backend (them typing their shit wrong) and that the system is going to need them to reset their password. Why would you even argue with them? Sometimes I’ll just say things like “happens to me sometimes too!!” Even if I’ve never been that dumb in my life.

Just make it work for them. Tell them it’s fine and it works now and to call you when it stops working again. “Happy to help!”

ExpensivePoint3972@reddit

Blame it on the very real phenomenon of cosmic rays changing bits from time to time.

How Cosmic Rays Affect Super Mario Speed Runners

RainStormLou@reddit

I usually say "in what way does this mean Satya Nadella can say some dumbass shit with a big smile at the next Microsoft press conference" or "does the vendor profit if this is a pain in the ass for paying customers" and we usually have our answer at this point.

Seriously though, if it's some goofy shit that's probably related to a "fix" Microsoft is deploying, I change "shit" to "stuff" or "thing" or "behavior" and send it lol.

dhardyuk@reddit

Well, computers used to occupy rooms the size of basketball courts and require a large 247/365 team just to cuddle them through the day. In fact the first computer bug was an actual massive insect shorting out connections in a computer.

I just tell users that the underlying technology is decades old, some bugs never get fixed c b because they are not worth the effort and as a consequence sometimes things take a while to sort themselves out.

That and turning it off and on again is how they cleared the original bug and it is still a good way to clear things down.

Also a restart or reboot is not the same as a shutdown and fresh start.

My own technique when I know things might want a little time to replicate is to advise the user to shutdown now, get a coffee or whatever and only start their computer when they have got back to their desk.

OneSeaworthiness7768@reddit

This is where soft skills come into play

mr_lab_rat@reddit

I like to use the wording “outside of our control”.

This often points the finger at the vendor and there is a chance they caused it.

When I’m stumped and need more time I just say that we did everything we could on this side with the user and I need to take more steps on the back end and/or with the vendor.

discgman@reddit

Microsoft updates work every time.

Excellent_Pilot_2969@reddit

This is a skill you learn in college...if you are a politics major

GardenWeasel67@reddit

I tell them "Welcome to the cloud".

Valdaraak@reddit

After getting through M365 SSO, they got stuck on the Teams login screen. Literally no button worked. After beating around the bush -- Incognito window, the works -- had to tell them to just try again later in like an hour.

In that case the answer would be "the issue was on Microsoft's end, likely due to delayed propagation between their backend servers."

Enough technobabble and plausibility (and in this case is probably pretty close to accurate) that they're satisfied enough to stop asking questions.

Do you blame the system, the user, or a glitchy universe?!

In that particular case, the latter. I've had similar issues before. No logged activity of a password change or reset anywhere, user verifies the password they're typing is 100% correct by clicking the little eyeball to unveil it, still says password wrong. Did a manual reset and got them back in.

The best glitchy universe explanation is to blame things on cosmic bit flipping (it's a thing). The most recent time I ran into that issue did happen to coincide with some pretty strong solar flare activity. It was a convenient explanation and a way to veer far enough off topic into what bit flipping is that they walk away learning something.

Every field and every industry has that kind of shit happen. Even a master mechanic might run into something in a car they've just never seen happen before and can't explain.

Hell, NASA has that happen. They just have rooms full of people dedicated to figuring out a) why that weird thing happened, b) what risks that implies, and c) how to never get caught by surprise by that specific thing ever again.

TheRedstoneScout@reddit

We've somehow managed to make our users and executives understand that Microsoft stuff is shit most of the time so when issues come up with MS apps they blame them.

Now when random shit happens on unrelated stuff we've gotten them to usually accept "technology is weird sometimes, let me know if it happens again"

A saying I've used for a while is "Once is a fluke, twice is a coincidence, three times is a pattern."

So, for 365/teams, that sync and provisioning of resources isn't instant on the backend. They appear to have witnessed what a half provisioned account looks like.

For okta and jira, that depends entirely on how they're built under the hood. None of the bigger webapps you see and use every day are some coherent, single, monolithic, executable running somewhere. They're a ton of interconnected components that can and often are run distributed across a bunch of separate servers. Your "hand data from A to B" has to go through networking, multiple intermediary services, etc. Some of those steps put it in more replicated, resilient, places, but if the right spot in the path "owns" that event/job/whatever at the right moment, and then hits a memory boundary or the like that triggers a reset of that service, it might throw your workflow or job trigger out the window before it gets where it's going. There IS an answer, but you're not going to find it looking at the shiny webpage.

User-centric cases: You don't argue it, but for explaining it to yourself... look up the Mandella effect. Human memory is hilariously inconsistent. Then watch an episode or three of House.

For the management questions on those things... some of it's just extremely technically complex systems and more variables than you're going to get an RCA for without investing a ton of resources. If it happens frequently, you rope in the vendor that you pay for support that is on the hook to know their system. If THEIR answer is "Dunno, happens sometimes. Can take about an hour to sync." you're completely covered. If they swear up and down it's instant, you get to start either proving them wrong by figuring out how to reproduce it... or you'll find your misconfiguration along the way. If management wants to expend the resources over a transient problem.