|
You’re gonna need to elaborate. What was the ssh error? Got graphs of the vm usage or preferably sar output. Changing a nic isn’t gonna be any help.
|
#
?
Jan 4, 2022 03:06
|
|
|
|
| # ? Jun 6, 2024 21:10 |
|
|
Or to be exact, it may theoretically help - I've had weird errors with broadcom nics that cleared up with an Intel replacement. But don't start by throwing hardware at the problem.
|
|
#
?
Jan 4, 2022 15:44
|
|
|
is there a reasonable way to launch a program in a local X session over SSH? Like, SSH into the server, and run the program, and have it pop up on the server? I don't generally need to interact with it, I can just kill/restart it, but it crashes alot and I don't want to deal with X forwarding it
|
|
#
?
Jan 4, 2022 20:28
|
|
|
You can do it by messing with $DISPLAY on the remote server but you'll have to use xhost in the server's x session to convince it to accept the connection. Once upon a time it was "okay" to run an x server completely open but all distributions close it down now because people got tired of idiots launching a million xeyes on their buddy's workstation (also it was a major vulnerability). The more modern way to do it would be run the app in a vnc server on the server. But if the app crashes constantly that doesn't buy you much.
|
|
#
?
Jan 4, 2022 20:33
|
|
|
I was afraid it was too locked down to be practical. ah well
|
|
#
?
Jan 4, 2022 20:52
|
|
|
Marinmo posted:While I agree with this sentiment (podman being superior to Docker), since it's the late comer to the party - and I'm currently experiencing this firsthand - it's a massive PITA to get to work properly compared to Docker, if you actually want to use the nifty features that podman brings (running rootless mainly, systemd integration secondly). Every single container out there (not really, but a large majority) expects to be run under rootful Docker, Linuxserver.io especially. Combine this with the fact that error logging is slightly ... Subpar, to put it nicely ... Makes for a very long and interesting ride should you attempt it. I'm currently moving over my entire Docker-stack to podman (on Fedora Server, so add in SELinux to the fun mix too!), consisting of the following: I did find a really neat and seamless way of enabling Wireguard for one or many containers, wg-pod. Syntax is dead simple, wg-pod join containername /path/to/wg.conf. You can use the -d switch to delete all other routes in the resulting netns and place it in an ExecStartPost in the service file. Caveat: it requires CAP_NET_ADMIN which means you'll have to fiddle with yet-another-permissions-system-in-Linix, luckily, here's your solution: # setcap cap_net_raw+eip /path/to/wg-pod (usually /usr/local/bin) - do note I'm not a 100 % the i is necessary here but I don't want to mess with it - works for me. If you use the same wg.conf for more than one container, wg-pod will join subsequent containers to the same netns. Just super convenient. As everything still is a bit of mishmash I really want to hold off on publishing anything about the process, but there's really something to be said for it all, that's for sure.
|
|
#
?
Jan 4, 2022 21:10
|
|
|
RFC2324 posted:is there a reasonable way to launch a program in a local X session over SSH? RFC2324 posted:I don't generally need to interact with it, I can just kill/restart it, but it crashes alot and I don't want to deal with X forwarding it
|
|
#
?
Jan 4, 2022 22:14
|
|
|
ExcessBLarg! posted:You want the program to run on the server and have it appear on the X session running on the server (the "X client"?). In that case all you should need to do is set DISPLAY and maybe XAUTHORITY. The xhost issue isn't a problem since it's still all running local to the server. its not the wholeprogram that crashes. its part of a stack that controls a bluetooth device, and if it receives an instruction while there is no device attached it stops allowing connections. Long term I plan on setting up handling so that the rest of the stack sees the device disconnect and won't accept commands during that time, but until then I was hoping for a cheap workaround.
|
|
#
?
Jan 4, 2022 22:43
|
|
|
I have a weird one, at least I can't figure out why it happens. I have a Windows install that has a couple of drives, actual physical drives. One of these drives contains my games. I also use a KVM VM to which I pass through this exact some drive, so that I can load my games from the drive when I don't want to dualboot. This is just a raw disk mapping, no fancy translation layers. Now whenever I load the Windows install again, it says that the drive that has been passed trough need checking. My question: Why does this happen? If I just shutdown the VM in a normal way, NTFS should be in a ok state right? (buffers get flushed, journal is taken care off etc.) Why would the other Windows install think it isn't in order? How would it know? I didn't know where to post this, but seeing that KVM is involved I thought I'd try it here first. Edit: AFAIK the disk does not contain any OS related things, just some folders for Steam and Origin. Mr Shiny Pants fucked around with this message at 10:54 on Jan 7, 2022 |
|
#
?
Jan 7, 2022 10:38
|
|
|
When you say raw device mapping, are you sure it isn't using virtio-blk over something like virtio-scsi? The first is plain block devices (with all the caching that that entails on Linux), while virtio-scsi is a proper SCSI implementation that fully conforms to the spec. It's also possible that Windows is just picking up on the controller changing enough that Windows thinks it warrants checking the disk. I don't know if KVM supports it, but I've found that the AHCI emulation in bhyve on FreeBSD avoids this whenever I've been doing something similar.
|
|
#
?
Jan 7, 2022 14:51
|
|
|
Odd SSH error: We have a security compliance scanner that uses SSH public keys to connect to various boxes to run various tasks. When connecting to a specific RHEL 7 box, it completes the key exchange and authenticates successfully, but then immediately after executing a command the connection is dropped. Jan 7 09:09:24 server sudo: our_user : TTY=pts/1 ; PWD=/home/our_user ; USER=root ; COMMAND=/bin/sh -c printf "command_start_%s" "nOMpQPKf"; netstat -a -n; printf "command_done_%s" "6qNTMuwy" Jan 7 09:09:24 server sshd[22959]: Read error from remote host 10.1.1.44 port 50620: Connection reset by peer The user executing the command has the appropriate sudo rights. Anyone have any idea what could be causing this? Increasing debug logging doesn't give us anymore insight, unfortunately.
|
|
#
?
Jan 7, 2022 16:59
|
|
|
BlankSystemDaemon posted:When you say raw device mapping, are you sure it isn't using virtio-blk over something like virtio-scsi? I wasn't clear, I am sorry. When the disk gets returned from KVM into the non VM Windows install, Windows will complain about the disk. For this install there is no controller changing or anything. Mr Shiny Pants fucked around with this message at 17:28 on Jan 7, 2022 |
|
#
?
Jan 7, 2022 17:02
|
|
|
Hypnobeard posted:Odd SSH error:
|
|
#
?
Jan 7, 2022 17:16
|
|
|
Pablo Bluth posted:As it's redhat, SELinux? Sorry, should have included: SELinux is (temporarily) disabled.
|
|
#
?
Jan 7, 2022 17:38
|
|
|
acetcx posted:A few days ago I started getting random freezes when gaming on my Fedora 35 system with an nvidia 2070. It sometimes works fine for an hour or two before freezing but sometimes it only take a few minutes. Only the video output freezes - the audio keeps playing and the system is still running so I can kill the offending game. Just as a follow up I'm pretty sure this was a hardware problem due to a DOCP overclock on the system memory. I forgot that I upgraded my RAM a few weeks before this started happening and I must not have played any games during that period. I had configured a DOCP overclock (from 2133 MT/s to 3200 MT/s) in the BIOS when I installed it figuring it would be safe enough. A few days ago I managed to get my hands on an AMD card and even with the new card the crashes continued so of course I stopped suspecting the nvidia driver. Some googling led me to a report somewhere that memory overclocks can cause GPU crashes. I've since turned off the overclock and it seems stable after a couple hours of gaming. Lesson learned, I thought XMP/DOCP overclocks were safe but apparently not always.
|
|
#
?
Jan 8, 2022 06:22
|
|
|
Hypnobeard posted:We have a security compliance scanner that uses SSH public keys to connect to various boxes to run various tasks. When connecting to a specific RHEL 7 box, it completes the key exchange and authenticates successfully, but then immediately after executing a command the connection is dropped.
|
|
#
?
Jan 8, 2022 07:06
|
|
|
Hypnobeard posted:Sorry, should have included: SELinux is (temporarily) disabled. Does a plain ssh execution sans command opens a stable connection? ..what if you then manually run the same command?
|
|
#
?
Jan 8, 2022 11:58
|
|
|
Through some vvvvs on it. You’ll figure it out
|
|
#
?
Jan 8, 2022 12:48
|
|
|
I have an interview for a Linux job coming up at one of the big cloud providers. What would you suggest I be studying? I have been using Linux in some capacity for a long time and I didn't have any problems with the technical assessments or the first interview. But these will be more in-person/virtual and less multiple-choice type stuff, and I'm sure I'll get drug into deeper waters of some things I'm not that familiar with. I know your basic LAMP stacks, troubleshooting, processes, how to read poo poo like top and and figure out if bottlenecks are in disk/cpu/memory, pretty good understanding of firewalls and dns/dhcp/http, have done a bunch of programming languages and can build crap from source and figure out those errors... I started this process in the midst of studying for RHCSA, which I feel pretty comfortable with but that's basically some memorization and zero real-world knowledge. My plan right now is just to binge on Linux-related YouTube videos. I just feel like I'm going to get marked down for not knowing some arcane command line option for something or some stupid file location. Now that I think about it, I haven't installed Arch in a few years...
|
|
#
?
Jan 10, 2022 02:33
|
|
|
Bob Morales posted:I have an interview for a Linux job coming up at one of the big cloud providers. Post job description
|
|
#
?
Jan 10, 2022 04:37
|
|
|
Methanar posted:Post job description BASIC QUALIFICATIONS The Linux role supports our services that focus on compute technologies. In this role, you will support AWS Services including EC2, EBS, LightSail and the services that enable customers to migrate their workloads to AWS. As well as supporting our customers, you will be leading operational improvement initiatives, an acting as the voice of the customer to internal teams across AWS. · Advanced experience with System Administration with Linux (Ubuntu, CentOS, RedHat) · Advanced of experience with Networking and troubleshooting (TCP/IP, DNS, routing, switching, firewalls, LAN/WAN, traceroute, iperf, dig, cURL or related). · In-depth understanding of virtualization and cloud computing (Hypervisors, VMware, Xen). · Bachelor’s degree in Information Science / Information Technology, Data Science, Computer Science, Engineering, Mathematics, Physics, or a related field OR equivalent experience in a technical position PREFERRED QUALIFICATIONS · Experience managing full application stacks from the OS up through custom applications · Knowledge of AWS and Cloud Computing concepts · Good understanding of security best practices. · Good understanding of distributed computing environments and methodologies. · Advanced degree in Computer Science or a related field.
|
|
#
?
Jan 10, 2022 05:22
|
|
|
Kinda amazed aws wants iperf knowledge. That’ll just show problems in their network.
|
|
#
?
Jan 10, 2022 05:36
|
|
|
Bob Morales posted:BASIC QUALIFICATIONS Brush up on whichever part of this you're weakest on. And try hosting an entire stack on AWS and seeing what the pain points are so you can talk about them.
|
|
#
?
Jan 10, 2022 06:49
|
|
|
Hypnobeard posted:Odd SSH error: Do you have Defaults requiretty in your /etc/sudoers or related config? RHEL/CentOS can get picky about what it requires from non-interactive logins and this might need to be tweaked. IIRC correctly it’s sort of outdated security advice anyway, so may well be something you can just remove.
|
|
#
?
Jan 10, 2022 08:07
|
|
|
Installing arch is pretty useless if they expect experience with Ubuntu, CentOS, RedHat
|
|
#
?
Jan 10, 2022 09:40
|
|
Associate Christ
|
other people posted:Installing arch is pretty useless if they expect experience with Ubuntu, CentOS, RedHat It's not especially useful if they have concrete practical questions, I agree. On the other hand, if he's going to a more general "are you the right sort of person" interview, I guess it could count positively? Showing interest in the greater Linux ecosystem and an ability to adapt to different ways to get to the same end result, and all that.
|
|
#
?
Jan 10, 2022 11:51
|
|
|
jaegerx posted:Kinda amazed aws wants iperf knowledge. That’ll just show problems in their network.
|
|
#
?
Jan 10, 2022 16:15
|
|
|
Hi goons, I've recently started to muck around with Linux distros on a few laptops. One of them I currently have Ubuntu DDE running. My problem is that I want to install another distro but my BIOS must be corrupt, as it doesn't let me change the boot order. I can get around this by opening up the laptop and deleting the partitions, but is there a way I can get the machine to reboot and pick up the USB install media?
|
|
#
?
Jan 11, 2022 00:13
|
|
|
UEFI? Can you enable and boot to the EFI shell? You can also try to use efibootmgr to reorder or remove boot entries from the NVRAM.
|
|
#
?
Jan 11, 2022 00:15
|
|
|
ExcessBLarg! posted:UEFI? Can you enable and boot to the EFI shell? I should have made myself more clear - I'm loving clueless about lots of this poo poo, I'm out of my Windows comfort zone. I don't know what any of that means! Edit 2: Right, I've gone through the instructions for changing the boot sequence. Running efibootmgr once I've done my edits shows the USB is the first boot entry, but when I restart the machine it just seems to boot straight from the SSD again. Running efibootmgr after a restart shows that the boot order has reverted. Am I missing a step? WattsvilleBlues fucked around with this message at 00:39 on Jan 11, 2022 |
|
#
?
Jan 11, 2022 00:22
|
|
|
It should all work that way but BIOSes are weird. Regarding the EFI Shell, it's usually an option to turn on in your BIOS configuration. You may have to disable Secure Boot first to see it. The EFI shell is a command prompt in your BIOS. If you're able to access it you can tell EFI exactly what to boot. Usually USB media will be something like: FS0:EFI\BOOT\BOOTX64.EFI FS1:EFI\BOOT\BOOTX64.EFI and so on.
|
|
#
?
Jan 11, 2022 01:22
|
|
|
WattsvilleBlues posted:I should have made myself more clear - I'm loving clueless about lots of this poo poo, I'm out of my Windows comfort zone. Can you get into the BIOS on your laptop? It's usually something like tapping Delete or sometimes a function key to access before the OS boots. If you're not sure and you're not getting a splash screen when you restart then just searching on your laptop model and something like BIOS access key should give it to you.
|
|
#
?
Jan 11, 2022 01:31
|
|
|
Also, which laptop? Laptop BIOSes are notoriously weird and fickle, so there may be some specific options or workarounds needed.
|
|
#
?
Jan 11, 2022 01:54
|
|
|
Even if you can't get into the bios, there might be a separate boot order menu. It usually is at f8 or or f11 or f12 and on my lenovo laptop it is behind a hardware button that needs a paperclip to reach. And I once used a computer with a corrupt bios for quite some time, it only lost the settings when I turned off the computer but kept settings for software reboots. There also should be a way to chainload your usb-stick from whatever boot manager you current distro uses, but I never worked with that one.
|
|
#
?
Jan 11, 2022 02:29
|
|
|
Anyone happens to know what's the deal with Samba and RDMA support (i.e. SMB Direct)? I'm looking it up like every so often, and it's still not implemented, yet I just ran across references of prototype code in 2014 of all things. Meanwhile, someone else comes along, casually shits out a kernel SMB driver than happens to support it just like that. And meanwhile Samba only updated the implementation plan/document again in October 2021.
|
|
#
?
Jan 13, 2022 00:55
|
|
|
I'm pretty sure the prototype code was found to be tainted (ie. the coder admitted to looking at leaked Microsoft code). It's much the same problem that the BSDs have if they want to avoid GPL code.
|
|
#
?
Jan 13, 2022 01:19
|
|
|
I am running KDE Neon on an XPS 13 9370. I have a POD Go (https://line6.com/podgo/) which is a guitar amp/effect modeler, but also acts as a class compliant USB audio interface. I am able to plug it in and record through my DAW (Bitwig) with ease, little to no configuration needed. What I cannot figure out how to do, though, is output audio from my laptop (Firefox audio, specifically) to the POD so that I can listen to what's playing in my browser at the same time I am playing guitar. Headphones are plugged into the POD. Oddly, if I record audio into a track in Bitwig, then I am able to play that track back simultaneously while playing and recording new audio on top of it. Bitwig seems to know how to output audio through the interface just fine, but I have no idea how to do this at the desktop level so I can push browser audio through. I have a hunch the answer is either "not possible" or "use jack," which is really intimidating. edit: I've discovered that using a Focusrite 2i2 can do this the way I need it to, without much fuss. I guess the issue is with the POD itself. meatpath fucked around with this message at 23:10 on Jan 14, 2022 |
|
#
?
Jan 14, 2022 22:04
|
|
|
random question thats always bugged me: why are symlink standard practice and not multiple hard links? is it just because its a pain in the rear end to make sure data is really deleted with a hard link, or is there some other reason?
|
|
#
?
Jan 14, 2022 23:56
|
|
|
RFC2324 posted:random question thats always bugged me: why are symlink standard practice and not multiple hard links? is it just because its a pain in the rear end to make sure data is really deleted with a hard link, or is there some other reason? At a guess: Hardlinks are implemented by the file system, so they can't point across mount boundaries - and not all filesystems support them. Softlinks are way more likely to succeed.
|
|
#
?
Jan 15, 2022 00:03
|
|
|
|
| # ? Jun 6, 2024 21:10 |
|
|
Hard links can't cross file systems and look like normal files (you have to check the link counts to spot them) so they confuse un-observant admins. The mantra of not using them kinda got cargo culted.
|
|
#
?
Jan 15, 2022 00:03
|
|