this post was submitted on 28 Feb 2024
835 points (97.4% liked)

linuxmemes

20880 readers
2 users here now

I use Arch btw


Sister communities:

Community rules

  1. Follow the site-wide rules and code of conduct
  2. Be civil
  3. Post Linux-related content
  4. No recent reposts

Please report posts and comments that break these rules!

founded 1 year ago
MODERATORS
 
top 50 comments
sorted by: hot top controversial new old
[–] [email protected] 86 points 8 months ago* (last edited 8 months ago)

Updates in Linux are far more tolerable. There’s really no reason to delay Debian stable, imo, unless you absolutely can’t risk some downtime.

Server rats excepted, it’s just a process that goes in the background and at most, you have to reboot the kernel.

There’s no staring at the Blue Screen of Boredom while windows update holds your machine hostage.

[–] [email protected] 66 points 8 months ago (6 children)

I work at a medium size company with hundreds of Linux servers and none of them get updated. Because it's more important that they keep running as they are than to have the latest updates. I bet this is very common for most companies.

[–] [email protected] 86 points 8 months ago (1 children)

There is nothing more important than security patches on a system.

I used to work at an FMI, which’s motto was “keep things stable”. Even the ciso department bought that crap. Until we hired a white hat hacker. The only thing given was the name of the company. He managed to get into the building, access an employee’s workstation and install a root kit on one of the most important financial message tracking systems (you know, the one that instructs other systems to transfer money), using a security bug, which would have been patched if they kept a regular (security) update cycle. After shit hit the fan, many people were fired and an update cycle was introduced.

No system is important enough to not patch. And if you believe it is, you’re wrong.

[–] [email protected] 9 points 8 months ago* (last edited 8 months ago) (5 children)

Yeah, but that just takes way too much work. You think I really care about the company's/bank's money if I'm not getting paid enough for that job? Security patches can also introduce new problems, like x changes, so y doesn't work, so the main app doesn't work... and what, then I have to manually edit code, introduce the thing that x relied on so that y can work again?

I'm sorry, but this is not your average IT department's job... or if it is, I expect a damn good compensation for it.

I've updated and rolled back snapshots because of shit like this... nah, not gonna try and figure out what the problem was... at least not for the salary I'm currently getting paid. If it burns, it burns, so be it.

[–] [email protected] 20 points 8 months ago (4 children)

I’d be surprised if you actually saw anything change from security updates tbh, I don’t think I’ve ever seen anything break from a quick patch.

Dist upgrades are when things might break, but they’re only once every few years. Leave them too long though and you may end up with compatibility issues if you need to make changes.

Fair enough if you’re not getting paid enough, the company should hire more people to stay on top of that though.

load more comments (4 replies)
[–] [email protected] 17 points 8 months ago (1 children)

God I hope I don't use any products from your company.

load more comments (1 replies)
[–] [email protected] 15 points 8 months ago (10 children)

“Way too much work” — if you ever said that where I work I’d fire you or not hire you in a heartbeat. An administrator’s role is not only to the stability of the system but the security too. You’re a hackers wet dream.

[–] [email protected] 10 points 8 months ago (1 children)

The phrase "Fuck you, pay me" comes to mind.

Cheapskates don't get top of the line security hardening. Pay more now or suffer a breach and pay contractors $1000/hr to fix your broken shit because you paid minimum wage for an administrator position and wanted them to do 5 jobs at once.

[–] [email protected] 7 points 8 months ago (6 children)

This guy gets it... and probably doesn't live in the US, cuz he knows the term "work 5 jobs at once".

load more comments (6 replies)
load more comments (9 replies)
load more comments (2 replies)
[–] targetx 26 points 8 months ago

If it's important that it keeps running then it should just be redundant and taking one node down for an update shouldn't be an issue. I know this is wishful thinking for a lot of services but I refuse to be on call for something if the client can't be bothered to make it redundant.

[–] [email protected] 16 points 8 months ago (1 children)

Jup same here. We have a colleague that constantly reminds everyone that we're not properly patched (even running eol versions) but there's always something to be done that's a higher priority.

[–] [email protected] 7 points 8 months ago (1 children)

Exactly. Shit needs to just work, period. Why? Because otherwise, I'm the one getting 2AM calls... and I would be OK with that if I'm properly compensated for it... which I'm not.

[–] [email protected] 4 points 8 months ago (1 children)

Did you think of testing security updates on a staging environment before going in production with it, if you suspect in can break things?

I think there is no excuse to apply security fixes wich have a CVE number.

If you are on Debian stable unattended updates are not a problem.

[–] [email protected] 4 points 8 months ago* (last edited 8 months ago) (2 children)

See, building and configuring a staging environment also takes time and money... money which they are not willing to spend on something "for testing" and not in actual use. Plus, I'm not gonna get paid for doing that either, so why actually do it... to be honest, I would do it, even for free, but you gotta caugh up the money for the hardware man. I've been told "just use what you have in the scrap pile"... for what, a server 🤨? Are you serious? They barely spend any money on that even, why should I bother creating something as e staging environment.

load more comments (2 replies)
[–] [email protected] 7 points 8 months ago

Not at all.

[–] [email protected] 6 points 8 months ago

Typically monthly or quarterly patching depending on severity and DMZ exposures. When log4j or shellshock hit it was patch once the patch was released and tested

[–] [email protected] 3 points 8 months ago (2 children)

If it's a personal server that can manage being down for 15min or so. You could just setup auto updates with email if anything goes wrong and reboot off hours. Containers also make it less risky although it does fail to update every once in a great while.

load more comments (2 replies)
[–] [email protected] 42 points 8 months ago (2 children)

Do you work for the North Korean government or something OP? Why discourage people from keeping their systems secure?

[–] [email protected] 31 points 8 months ago (2 children)

What they are referring to is people just don't update their server because during that time they wouldn't be able to make a profit. This goes more to middle siszed businesses but happens rather often

[–] [email protected] 9 points 8 months ago (2 children)

Blows my mind, lol. Usually means no redundancy that allows one set to be done while the other set handles the traffic.

[–] [email protected] 9 points 8 months ago

"Why should we pay for another server one works just fine, a second would just be waisted money."

Also

"We need 9 9s of reliability or the company will fail."

[–] [email protected] 7 points 8 months ago (5 children)

Yeah, it is quite common, I can confirm... well, at least around here it is.

load more comments (5 replies)
[–] [email protected] 5 points 8 months ago (2 children)
[–] [email protected] 15 points 8 months ago

Joke transfer unsuccessful. Server crashed. Time to update the joke server.

load more comments (1 replies)
[–] [email protected] 9 points 8 months ago

Security is an art... the art of not giving a fuck about your data

-Op, probably

[–] [email protected] 30 points 8 months ago

I find this to be least acurate with debian.. on other distros a patch may or may not install a new version of that package. that can bring changes to the behavior. On debian stable the security issues are backported. So you can patch and be sure that there is no changes to the behavior of the system. It is basically the reason all vm's i manage are debian stable.
It is also true they never crash. But that is expected of linux. It is the extreme reliabillity that is the debian killer feature for me.

[–] [email protected] 21 points 8 months ago (1 children)

Me with my 'homelab' nas:

system (user-facing) package has an update? It'll auto-update overnight

dockerized service has feature updates? Let watchtower handle it with the weekly schedule

dockerized service with security patch? yeah, let's hit that this afternoon

actual system update? EVERYTHING IS GOING OFFLINE -4 SECONDS AGO FOR THIS

[–] [email protected] 19 points 8 months ago (1 children)

The system is going down NOW.

[–] [email protected] 19 points 8 months ago

Debian updates are not usually that big of a deal especially if you have HA configured

[–] [email protected] 12 points 8 months ago (1 children)

i'm pretty sure security updates are optional.

[–] [email protected] 5 points 8 months ago

Just put a "these colors don't run" text in the log in

[–] [email protected] 12 points 8 months ago (5 children)

Unatended-upgrades keeps all systems securly patched. But there is a need for a reboot for kernel updates now and then.

load more comments (5 replies)
[–] [email protected] 11 points 8 months ago (1 children)

True except for the one BOFH admin on the team who actually cares about best practices.

And yes, most distros have painless updates, the devs and everyone else don't care.

[–] [email protected] 13 points 8 months ago (1 children)

Hi. It's me. The guy bitching about best practices every other meeting. Sorry, but some of my past and present coworkers are clowns.

load more comments (1 replies)
[–] [email protected] 10 points 8 months ago

"Until you crash, no on ~~cares~~ will reboot you."

[–] [email protected] 6 points 8 months ago (1 children)
  • until there's a PCI audit.
[–] rushaction 2 points 8 months ago

I have two words for you, "compensating controls."

It's like goddamn magic.

[–] [email protected] 4 points 8 months ago

yes, im guilty of this. haven't got time to update my server to v12

[–] [email protected] 3 points 8 months ago

Isn't live patching a thing?

load more comments
view more: next ›