@ilkkachu come on, please don't drag out irrelevant conversations in the comments. I told the OP that they should never run commands blindly without knowing what they are. Of course that applies to cat as well, I never said otherwise!

(re unix.stackexchange.com/questions/669981/…)

But, also obviously, it is much easier to misunderstand "dd if=/dev/null of=/dev/sda" than "cat /dev/null > /dev/sda" since the 'if' an 'of' are very similar and only differ in one letter.

I am not saying that dd is inherently more dangerous than cat. Hell, all I was saying is that you shouldn't blindly run commands!

surely nobody would run anything they don't understand... :P

And I think sudo cat doesn't work like that due to the redirect, so you have to work a bit more for that. Of course you can still overwrite your ~.

yep

or run as root

@terdon No, that's exactly what you were saying: "Especially if that command is dd, a very powerful and unsophisticated tool" and "even more important with something like dd that can be missused so easily." And exactly what I was objecting to.

@ilkkachu Yes. That is what I am saying if you completely ignore the first sentence which is in italic bold font for emphasis. Sure, if you take something out of context and fixate on it, and then also ignore the explanation I gave above, you can read it that way.

But even if you do read it that way, if you honestly cannot see that running 'sudo dd if=/dev/null of=/dev/sda' is much easier to happen by mistake than running 'cat /dev/null | sudo tee /dev/sda', then I don't really know what to say and we can agree to disagree. As far as I'm concerned, the syntax of dd makes it orders of magnitude easier to missuse.

@terdon "ignore the explanation"? I didn't see anything in the comments backing up the claim that dd would be more dangerous than any other command. Those quotes were from the comments, from the part I objected to. Yes, you said it here that you think if and of are similar, and ok, perhaps they can be confusing (if one doesn't realize the connection to "in" and "out"). But like I said, < and > are also just one character apart, and on the same or neighboring keys on many keymaps too.

@terdon there was no sudo there.

@ilkkachu That's probably because I never claimed that dd is more dangerous.

I said "don't run commands you don't know". And then I said that is especially applicable to dd. Not because it is more dangerous (that's how you chose to read it) but because the syntax is obscure ("unsophisticated tool") and very easy to get wrong.

@terdon well, you twice pulled it as an example of a particularly dangerous command, so yeah, one might get the impression you think it is somewhat more dangerous than a random other command, no?

@ilkkachu Yeah, OK. I'll ask another mod to clean up the comments. Next time, please take this sort of thing to chat instead of continuing in the comments.

@terdon yeah ok. What I saw there, was that you answered my comment by basically just repeating what you said the first time. Instead of giving any new information, rephrasing your point in some other way, or moving to chat immediately. But I guess that's fine, since you can ask someone to just delete the comments?

@terdon also, I do honestly think the biggest issue with commands like that is mistyping the target partition. It doesn't matter if the command there is dd or cat or tee or fdisk or wipefs, if one types the wrong device name, then bam. They're usually one character away.

@ilkkachu Yes, I repeated that you shouldn't run commands if you don't know what they do. And of course that is even more true of commands like dd than commands like echo.

@ilkkachu exactly

which is why you shouldn't run them blndly and then come here to ask what they do.

@terdon well, at least we agree on that part :D

The biggest mistake is not running commands one doesn’t understand, it’s not having up-to-date backups which can easily be restored ;-).

Although running commands one doesn’t understand can have bad consequences other than losing data, so I guess that is really the bigger mistake here.

@StephenKitt ooh! well said

(Recent instances here which resulted in users installing cryptominers spring to mind.)

I was wondering if you meant something like that or actually breaking stuff. What was the thing about a filesystem interface to some EFI parameters was left read-write, and then you could trash the EFI settings with an rm -rf ...

I feel like asking a question, is ZFS safe to use on Debian, or not? Because I hear contradictory stuff.

this was the EFI+rm fiasco: lwn.net/Articles/674940

My current VM is OpenVZ, which according to my service provider, does not support upgrades across Debian versions, I think because the kernel cannot be upgraded.

But I think the other options might be more expensive.

I still see those mounted and available on my system

terdon@tpad ~ $ ls //sys/firmware/efi/efivars/ | wc
    158     158    7624
terdon@tpad ~ $ ls -l //sys/firmware/efi/efivars/ | head
total 0
-rw-r--r-- 1 root root   12 Sep 18 13:53 AppName-1fd8b79f-0be2-4d57-b241-81c5e24e01a1
-rw-r--r-- 1 root root   12 Sep 18 13:53 AppPlatform-1fd8b79f-0be2-4d57-b241-81c5e24e01a1
-rw-r--r-- 1 root root    5 Sep 18 13:53 AuthVarKeyDatabase-aaf32c78-947b-439a-a180-2e144ec37792
-rw-r--r-- 1 root root  304 Sep 18 13:53 Boot0000-8be4df61-93ca-11d2-aa0d-00e098032b8c

@FaheemMitha you won’t be able to use ZFS with an OpenVZ-based system

@StephenKitt No, ZFS would be for my personal machine. Or machines. Not for a VM.

Assuming I try it. I hear contradictory things.

@FaheemMitha apologies, I thought your two comments were related.

@StephenKitt They were unrelated. I was just musing about ZFS.

I do need to upgrade my VM though. It's currently still on Debian 8.

Which is becoming a problem.

Just trying to understand my options.

I'm using RamNode, which has worked well for years. And is cheap.

Is KVM better than OpenVZ? They seem to be offering more than I'm currently paying for my OpenVZ. At ramnode.com/index.html

Unless I'm missing something?

KVM gives you a full VM with its own kernel, rather than a VE in OpenVZ with a shared (very old) kernel.

The big limitation with OpenVZ now is that its kernel is too old, and can’t run systemd.

Maybe it's that they are only offering 35 GB disk space SSD (if I understand correctly).

That seems very small.

@StephenKitt So that is better, because I can upgrade without reinstalling.

The OpenVZ plan is giving 150GB. Which is still not a lot, really.

Can I run Debian 11 on an OpenVZ, or is that not even an option?

@FaheemMitha you can upgrade on OpenVZ without reinstalling, as long as you avoid the kernel-related limitations

@FaheemMitha I’m not sure whether Debian 11 supports running without systemd

@StephenKitt I don't see how I can avoid the kernel-related limitations. The last time I tried to upgrade, it errored out, and RamNode support informed me what such an upgrade was not possible.

Otherwise, I would not still be on Debian 8.

Can anyone confirm if I understood the 35 GB disk space correctly?

@StephenKitt Yes, perhaps it's not possible.

@terdon oh, I don't know exactly what's happened with that. The LWN article mentions making them immutable (chattr +i), so a plain rm even as root can't delete them. I always assumed it was fixed in one way or the other, since it's a rather bad failure mode, but I don't know what exactly happened in what kernel (or systemd) versions. Or didn't happen.

@ilkkachu Ah. I have no idea how to check for that

terdon@tpad foo $ sudo chattr +i file
terdon@tpad foo $ ls -l file
-rw-r--r-- 1 terdon terdon 167 Sep 21 18:54 file
terdon@tpad foo $ cat > file
bash: file: Operation not permitted
terdon@tpad foo $ rm file
rm: cannot remove 'file': Operation not permitted
terdon@tpad foo $ stat file
  File: file
  Size: 167       	Blocks: 8          IO Block: 4096   regular file
Device: 806h/2054d	Inode: 12876298    Links: 1
Access: (0644/-rw-r--r--)  Uid: ( 1000/  terdon)   Gid: ( 1000/  terdon)
Access: 2021-09-21 18:54:40.950576752 +0300

@terdon lsattr

it's a totally Linux thing, AFAIK. Not sure if even all the Linux filesystems support it.

(as in like even tmpfs doesn't seem to)

/sys/firmware/efi/efivars$ lsattr
----i--------------- ./SystemAccess-e770bb69-bcb4-4d04-9e97-23ff9456feac
----i--------------- ./HpPassphraseStructureVariable-707c9176-a4c1-4e27-851d-1c37b7ca73c8
...

@ilkkachu ah, duh, thanks

And yes, it's set for the efit vars

Most of them anyway

$ lsattr //sys/firmware/efi/efivars/ | cut -d' ' -f1 | sort | uniq -c
     35 ----------------------
    123 ----i-----------------

yeah, I can see some like BootCurrent, Con{In,Out}Dev, PlatformLang, Timeout that don't have it. All of those sound rather non-dangerous.

lol, famous last words ("sound" non-dangerous)

But yeah, that's what I'm assuming too

@terdon hahah, yeah, I'm not going to try messing with them, or recommend anyone else to do that without actually finding out either :D

Does anyone have any VM recommendations? I seem to recall there was a site that made such recommendations.

@FaheemMitha What do you mean, like VirtualBox vs. VMware?

@Kusalananda Sorry, I was very unclear. I meant a VM service, sorry. I guess VPS would be a more accurate term. I.e. Virtual Private Server.

Ah, ok. No, I know nothing about things like that. Sorry.

@Kusalananda OK. I figure not everyone uses such things.

back in my day they called it "in the cloud"!

@FaheemMitha There are dozens. I found mine by searching for "VPN review".

@AndrasDeak They still call it that.

@terdon I'd prefer something in the USD 5 range. Like my current one.

@FaheemMitha if you would like to indirectly support Debian, check out Leaseweb

@StephenKitt OK. Thank you for the reference.

The disk space is really small, though.

I’ve never used them so I don’t know anything beyond the fact that they provide discounts to members of the Debian project.

Maybe a dedicated Borg setup.

@FaheemMitha if you want that, check out rsync.net

Though then I wouldn't have any control over the remote.

@StephenKitt I remember looking at that previously.

The trouble is then I won't have a dedicated VPS if I need one.

I wonder why the allocated disk space is so small.

I’ve used OVH VPSs in the past, they’re OK.

@FaheemMitha yes, that's what search engines and review sites are for.

They all seem to offer very small amounts of disk space at that price. Even Borg can't perform miracles.

Well, relatively small. Once upon a time 25GB was a lot of space.

@JeffSchaller Quite so.

@JeffSchaller this.

@StephenKitt Was your warning about systemd and OpenVZ because of a bug in systemd? A quick search suggests something along those lines.

And if there is a bug, why hasn't it been fixed?

> The big limitation with OpenVZ now is that its kernel is too old, and can’t run systemd.

@FaheemMitha depends a bit on what you do. I think I've seen e.g. Raspberry Pi kits with a 16 GB SD card, and I think that wouldbe plenty for a lot of things.

@FaheemMitha systemd needs a 2.6.39 kernel or later, OpenVZ stopped at 2.6.32

wow

Huh, bugs.openvz.org/browse/… seems old but talks about fixing at least one bug: "fixed systemd v220+ failure inside container (OVZ #2693)"

@StephenKitt neat

@ilkkachu also neat :P

@StephenKitt Oh. Bummer.

@ilkkachu In part, backup. Actually, in practice almost entirely backup, but it would be nice to be able to do other things.

So the OpenVZ project is dead, then?

Don't know anything about it, even though I was using it.

@FaheemMitha apparently not: wiki.openvz.org/News

It’s also possible the systemd issues have been fixed, as suggested by Jeff.

Unfortunately some VE setups aren’t particularly well maintained by their hosts and still use very old versions...

@StephenKitt Yes, I was just looking at that. But if not, why stop at 2.6.32? I think I read somewhere they were in the kernel and are now removed. Searching for more information.

wiki.debian.org/OpenVz

My mistake. Looks like it was never in the kernel.

@StephenKitt VE?

@FaheemMitha virtual environments

How bad is it to not use systemd as init? I've obviously never tried, since it's the Debian default, and I'm not, usually, overly masochistic.

@StephenKitt Thank you.

@FaheemMitha there’s a variant of vzctl which uses in-kernel features, same as LXC: wiki.openvz.org/Vzctl_for_upstream_kernel

@StephenKitt Sounds very new. I wonder if any hosting provider uses it. I'll search.

@FaheemMitha LXC has been around for a while, and is used by some hosting providers

@StephenKitt Ok, I'll search for that too.

Looks like Debian before v8 didn't use systemd as init. Not sure how long you've had Debian installed.

@JeffSchaller My VPS is currently/still Debian 8.

My home machine is currently up to date with Debian 11.

I love V8s, no replacement for displacement

Though it's getting long in the tooth, I really need to get a new machine before this one falls over without warning one day.

@jesse_b Huh?

@FaheemMitha engines

@StephenKitt Ah

@jesse_b "I could've had a V-8!"

nice

Stay a while and listen

anyone getting d2 resurrected?

d2?

diablo 2

Hi @Braiam. How goes it?