If he doesn't have root on the machine then it's highly unlikely that anyone is going to give him console access. Besides, to get console access, he'd need root edit and update grub. Your answer doesn't work.

@NasirRiley IP console access?

He can go into the BIOS and everything with that.

Disclaimer: I work for a CSP (Cloud Service Provider)

And a ton of people don't know that they can get console access.

Because they don't know it exists.

@NasirRiley If you're adamant about me deleting the A and closing as "non-repro" I'm willing to do that.

Deleted! Thanks for keeping on pushing me...

I need to go to sleep!

Good night all!

P.S. It's possible to boot an ISO file through PXE, bit not going to go down that road...

Neither would those suggestions work, either. He doesn't have physical access to the box so he can't get into the BIOS not can he PXE boot.

Out of curiosity....when my Lenovo dies, synapitcs is completely dead until I dual boot Windows. Anyone familiar with that?

@Fabby Just FYI, you can link directly to comments.

@NasirRiley That's what console access gives you: hardware control.

@FaheemMitha I know... Just informal discussion between £Nasir and me.

@RuiFRibeiro It's called a side effect.

no one has ever tested the Linux drivers on a totally dead system so one of the init parameters is not called correctly so the hardware remains dead.

@RuiFRibeiro You should file a bug on how to reproduce.

(I've had the opposite a few times over on AU: if warm booting from Windows in Ubuntu: Kaput, if cold booting: works.)

@RuiFRibeiro which Lenovo? The T480s touchpad is known to be broken in 4.18, I don’t know about other models. But I do know the Red Hatters who work with the Synaptics people to fix these issues...

Am I going slightly mad? ssh-keygen should create a new key each time it's run, shouldn't it?

I can't see anything explictly saying that, but it makes sense to me.

@TRiG excellent reference, and yes, I think it should

the first thing it does is ask what file to save the key in

@StephenKitt It doesn't seem to be, which is confusing. It offered to overwrite the old key, but the new key appears to be the same.

And then we deleted the files and tried again, and it still looks the same.

@TRiG right that’s very surprising

I just checked and I get a different key every time

how are you comparing the keys?

@StephenKitt I didn't actually check what the original key was, but the new one could still connect to Bitbucket, so I assume it's the same.

@TRiG ah, if you’re using an SSH agent, that will still have the old key

And then I did a cat of id_rsa.pub, deleted everything in ~/.ssh, ran ssh-keygen, did cat again and compared by eye.

So restarting the agent, or logging in and out again, should pick up a new key?

you need to run ssh-add -D before trying to connect to Bitbucket

@TRiG or that, yes

I'm wanting it to fail. Basically, I'm trying to disconnect the machine from one Bitbucket account and connect it to another.

Okay. Shall try that.

Thanks.

@StephenKitt Thinkpad E560. It works if you boot Windows first; then you can boot Linux as many times as you want as long the battery does not die on you. Then you have to boot up Windows again for it not to be dead. Some initialization code buried in the windows driver...

@RuiFRibeiro which version of the kernel are you using?

@Fabby I rarely file bugs without understanding better at technical level what is happening.

@StephenKitt 4.10.5 Debian/Antix

@Trig In Debian I delete the keys, and do dpkg-reconfigure openssh-server

@TRiG hmmm wrong tag.

@RuiFRibeiro for a client key?

@StephenKitt hmmmm.....I though he was talking about a server key. What is the difficulty creating client keys?

@StephenKitt And it worked.

@RuiFRibeiro Ubuntu, by the way.

@RuiFRibeiro the agent remembers the old key

@StephenKitt Any references were unintentional.

@StephenKitt hmmm, should you not use ssh-add then?

@RuiFRibeiro OK, it would be useful to test newer kernels; the old Synaptics driver used PS/2 extensions to talk to the trackpad, instead of the trackpad’s native mode (which is what the Windows driver uses); newer kernels (since around September 2017 IIRC) use the trackpad’s native mode

@RuiFRibeiro yes, we discussed that already

@StephenKitt makes sense. I just often try to stick to stock kernels, but then I am used to VMs and not to physical machines.

@RuiFRibeiro I don’t know what Antix is like but Debian provides recent “stock” kernels for the stable distro

@StephenKitt Antix is basically Devuan done right.

simplifying a lot...

@RuiFRibeiro I got that from your comments but that doesn’t tell me how they manage their kernels ;-)

@Fabby i'm sorry. i didn't mean to just disappear yesterday. i actually wrote 3 more text messages before i noticed that my internet connection went down. it didn't come back up again before i went to sleep

@StephenKitt Gonna check it better, thanks.

i think most of my activity is on askubuntu. however, i'm neither french nor russion

Leaving, see you all later on.

it's "chrstoph", not "christophe", so not french

it's a greek name, btw

i'm german

@UTF-8 No worries!

@RuiFRibeiro there’s 4.18.4 in the Antix package pool

@UTF-8 Ah, Noch ein Deutscher!

Anyway. in the meantime your question was closed as the wrong dupe.

(haven't had another look since yesterday though)

Glad it works!

@RuiFRibeiro It's a side-effect and notoriously difficult to debug.

So someone with the right ability and understanding is what's needed...

;-)

@Fabby @StephenKitt I confirm the bug is still present with 4.18.4

@StephenKitt just took out the battery after installing 4.18.4.

@RuiFRibeiro I wonder what Chromium has against your avatar. Firefox and Gimp are both OK with it...

Oh! Maybe it's because it's being loaded from Facebook. Probably being blocked...

@RuiFRibeiro it would be worth reporting to linux-input@vger.kernel.org

@StephenKitt I was wondering, do you know what happens if there already are $pid_max processess running so no more PIDs can be assigned? Does the kernel simply refuse to launch new processes? Is that because the kernel can't deal with more or simply because it has no more names to give them? I mean, could it theoretically launch more but it can't because it would have no way of tracking them?

@terdon You mean simultaneous? Otherwise it just wraps.

(seen this on our servers, comfirmed by Googling)

@terdon the fork call fails with EAGAIN

@Fabby Yes, simultaneous

and yes, the kernel could theoretically start more processes but can’t track them

otherwise you get ENOMEM

see man fork

$ errno -ls | grep -E 'EAG|ENOMEM'
EAGAIN 11 Resource temporarily unavailable
ENOMEM 12 Cannot allocate memory

I see.

Thanks.

@derobert Yeah, I have noticed once in a while problems with that.

@StephenKitt Maybe. If in the meanwhile I do not find more about it, I will try to do a better bug report.

@UTF-8 Here we use multiple net connections.

Well, two. The one I was using went out this afternoon/evening, and I had to switch.

@RuiFRibeiro You might want to upload a copy to SE, just to reduce the amount of tracking Facebook can do, if nothing else

Hi @derobert.

@derobert will have a look, just did it as convenient.

@StephenKitt And managed to solve it with a kernel parameter ;)

Gotta write a good question here with the answer.