Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
Mr. Crow
May 22, 2008

Snap City mayor for life

Mr Shiny Pants posted:

Hey guys, I am working on installing Wireguard and I am coming up blank.

Due to wiregueard not running on PFSense, at least without some hackery, I want to install it on my Linux machine. The issue is that I now have two gateways, one for the VPN clients and one for everything else. If I configure PFsense with another gateway in it's configuration it will send ICMP redirects to the clients looking for the VPN network and this works except when I connect from the VPN network directly to a machine in the network. It seems that they don't know "the way back" so to speak without adding some static routes. I really don't want to add static routes. How do I get my clients to know about the VPN network? I've been looking at RIP and other stuff.

My PFsense machine runs on my Linux machine as a VM and I was thinking of adding the WG0 interface to a private network, add a third interface to PFSense and have it do the routing via an Opt interface. But the WG0 interface does not like to be added to a bridge or something else.

Can I create a VM like private network inside Linux, add the WG0 interface to it, add a second Nic to it from Libvirt and route everything? There is a whole load off Linux networking stuff and I am afraid I am not seeing the forest for the trees anymore.

Just because you can doesn't mean you should. That's a whole lot of complication in your network stack and it's gaining you.... What?

Adbot
ADBOT LOVES YOU

BlankSystemDaemon
Mar 13, 2009



Mr. Crow posted:

Just because you can doesn't mean you should. That's a whole lot of complication in your network stack and it's gaining you.... What?
But that's the whole Linux experience!

CaptainSarcastic
Jul 6, 2013



Hey, I'm finally building a new desktop after 10 years and wanted to get a little input. It's based around a Ryzen 5 3600X on an X570 motherboard. I run openSUSE as my primary OS, and just have Windows installed for games or the odd thing that requires it (and at this point dual-booting is a really ingrained habit that I don't see changing).

My question is whether I should give Tumbleweed a try or stick with the point releases like I have done up to now?

The hardware is all new enough I am thinking Tumbleweed might actually be more likely to play nicely with it, and from what I've read the proprietary Nvidia support in Tumbleweed is pretty decent nowadays. Any significant gotchas from Tumbleweed I should be aware of?

Doing the point updates has been fine for me under openSUSE, so while the rolling release model would be nice (I used to run a rolling release distro years ago) it isn't a major concern, either.

So, opinions on runningTumbleweed versus the point release as a daily driver?

Hollow Talk
Feb 2, 2014
Use Tumbleweed, never look back. I've been using Tumbleweed since it's inception (and factory before that) on my private laptop and is been working well. Well enough, in fact, that when the time came to put Linux in my work laptop, I went with Tumbleweed over Leap, which had been running without problems for the past two years. :pseudo:

The Human Cow
May 24, 2004

hurry up
Does anybody want to help me figure out why a client's Red Hat server won't boot? It's an HP ProLiant ML110 G7. At the beginning of the week, it wouldn't boot or POST at all. Over the course of the week, we've replaced the memory, motherboard, and power supply. It was running two 72 GB SAS drives connected to a HP SmartArray P410 RAID controller with a 256MB memory module. When the controller was connected to the new motherboard, the computer still wouldn't boot, so we ordered a replacement. It arrived today, but appears to be DOA because the motherboard won't detect it at all. The client was getting antsy about downtime, so I grabbed an LSI RAID controller from an old server, installed it, and was pleased to see the Red Hat boot screen after a reboot. However, I'm greeted with this screen when it's trying to boot:
code:
Warning: pci_mmcfg_init marking 256MB space uncacheable.
Red Hat nash version 5.1.19.6 starting
Reading all physical volumes.  This may take a while...
No volume groups found
Volume group "VolGroup00" not found
Unable to access resume device (/dev/VolGroup00/LogVol01)
mount: could not find filesystem '/dev/root'
setuproot: moving /dev failed: no such file or directory
setuproot: error mounting /proc: No such file or directory
setuproot: error mounting /sys: No such file or directory
switchroot: mount failed: No such file or directory
Kernel panic - not syncing: Attempted to kill init!

Kernel alive
kernel direct mapping tables up to 20bfff000 @ 10000-1a000
I'm assuming that the RAID controller is seeing the drives correctly because it's trying to boot into Red Hat, but I don't know where to go from here. Please help! I won't be able to get back in there until Monday morning, so I'm trying to figure out what my options are going to be when I get there.

other people
Jun 27, 2004
Associate Christ
RHEL5? The initrd lvm conf can't find the physical volumes that the volume group with rootfs is composed of.

Using a rhel5 install media you can use rescue mode to inspect the initrd files and edit/rebuild as needed.

Hopefully you can read this as it gives some basic investigative steps: https://access.redhat.com/solutions/27731

Make backups!

CaptainSarcastic
Jul 6, 2013



Hollow Talk posted:

Use Tumbleweed, never look back. I've been using Tumbleweed since it's inception (and factory before that) on my private laptop and is been working well. Well enough, in fact, that when the time came to put Linux in my work laptop, I went with Tumbleweed over Leap, which had been running without problems for the past two years. :pseudo:

Cool. Thank you. I was leaning towards Tumbleweed but it's good to hear feedback from someone who has been running it.

Varkk
Apr 17, 2004

What is the current state of the proprietary nVidia driver vs the Nouveau driver? I have a Fedora 32 PC I have setup for the kids. Currently it has an i5-3xxx CPU and no external video card. I have been given an i7-4xxx with a GeForce 645 card. It will obviously be a good improvement for them. Currently used for some school work and light gaming, Minecraft and Portal mostly. Tried Kerbal Space Program but that was too slow.
Current plan is to just pop the SSD into the new machine just not sure whether to install the proprietary driver or not.

The Human Cow
May 24, 2004

hurry up

other people posted:

RHEL5? The initrd lvm conf can't find the physical volumes that the volume group with rootfs is composed of.

Using a rhel5 install media you can use rescue mode to inspect the initrd files and edit/rebuild as needed.

Hopefully you can read this as it gives some basic investigative steps: https://access.redhat.com/solutions/27731

Make backups!

I can't see that because I don't have a subscription, but I was pointed to this on Server Fault: https://wiki.centos.org/TipsAndTricks/CreateNewInitrd. Does that look like it's a good idea?

I agree that backups would be a huge help...I just got the call when things went south. A decent backup will be objective #1 after things are back up and running.

Thank you for your help!

Volguus
Mar 3, 2009

Varkk posted:

What is the current state of the proprietary nVidia driver vs the Nouveau driver? I have a Fedora 32 PC I have setup for the kids. Currently it has an i5-3xxx CPU and no external video card. I have been given an i7-4xxx with a GeForce 645 card. It will obviously be a good improvement for them. Currently used for some school work and light gaming, Minecraft and Portal mostly. Tried Kerbal Space Program but that was too slow.
Current plan is to just pop the SSD into the new machine just not sure whether to install the proprietary driver or not.

There is no reason not to go with the proprietary driver if you can (unless you're one of those free-software purists). With repositories available like rpmfusion or negativo17 it's usually just an "install and forget" procedure. Pick one and don't mix. Depending on game one can get equal or higher FPS in games than windows.

Varkk
Apr 17, 2004

That was my thought. I just wanted to confirm. My main PC is a Fedora 32 machine with a GTX 1050 using the RPMfusion repo. Just wanted to confirm before jumping in on the kids PC especially since it is an older card.

BrainDance
May 8, 2007

Disco all night long!

the libdns-export1110 package in Debian Sid has had an outstanding bug for about a month now, I've just been holding it and going on with my life.

code:
critical bugs of libdns-export1110 (&#8594; 1:9.11.18+dfsg-1) <Outstanding>
 b1 - #954736 - Upgrade to 9.16.1-1 causes dhcpd to die with SIGABRT
Summary:
 libdns-export1110(1 bug)
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=954736

From my limited understanding, it looks pretty serious. I don't want to install it and have DNS suddenly dead. But if its just been sitting there for a month, and Debian really wants me to install it (this is a new package, not an upgrade of an older one), maybe I should just go with it. Could someone take a look at it and tell me what the practical effects of dhcpd dying would be?

Double Punctuation
Dec 30, 2009

Ships were made for sinking;
Whiskey made for drinking;
If we were made of cellophane
We'd all get stinking drunk much faster!
That’s the DHCP server, not the client.

CaptainSarcastic
Jul 6, 2013



Varkk posted:

What is the current state of the proprietary nVidia driver vs the Nouveau driver? I have a Fedora 32 PC I have setup for the kids. Currently it has an i5-3xxx CPU and no external video card. I have been given an i7-4xxx with a GeForce 645 card. It will obviously be a good improvement for them. Currently used for some school work and light gaming, Minecraft and Portal mostly. Tried Kerbal Space Program but that was too slow.
Current plan is to just pop the SSD into the new machine just not sure whether to install the proprietary driver or not.

The only glitches I've had with the proprietary driver usually involved a kernel update or driver update being out of step with the other, and X not being able to launch, requiring tinkering. That hasn't happened in quite a while, though. OpenSUSE has an Nvidia repository that usually makes the process go smoothly.

Varkk
Apr 17, 2004

That is why I use the akmod version of the driver. It rebuilds the driver to when the kernel is updated. I have sometimes seen the kmod version lag a new kernel by up to a few days in the past.
I was mostly concerned about any issues with Wayland not being supported by the proprietary driver but it turns out the spin I have installed on that PC uses X11 by default so it doesn't matter.

Kirov
May 4, 2006

The Human Cow posted:

I agree that backups would be a huge help...I just got the call when things went south. A decent backup will be objective #1 after things are back up and running.
Boot the machine with a live cd or usb stick, and dump all filesystems before rebuilding or touching anything. Could you possibly install a comparable Centos VM, and restore the backups there? That gen7 is already on a borrowed time.

other people
Jun 27, 2004
Associate Christ

The Human Cow posted:

I can't see that because I don't have a subscription, but I was pointed to this on Server Fault: https://wiki.centos.org/TipsAndTricks/CreateNewInitrd. Does that look like it's a good idea?

I agree that backups would be a huge help...I just got the call when things went south. A decent backup will be objective #1 after things are back up and running.

Thank you for your help!

Simply creating a new initrd probably will not help until you have the lvm fixed up. And you won't be able to chroot from rescue mode because nothing can use those LVM volumes until the configuration is sorted out; there isn't anything to chroot into.

And I was really referring to backing up the existing initrd before you try making a new one!

I would guess that in rescue mode after a vgscan you'll see missing physical volumes and uuids not found.

Saukkis
May 16, 2003

Unless I'm on the inside curve pointing straight at oncoming traffic the high beams stay on and I laugh at your puny protest flashes.
I am Most Important Man. Most Important Man in the World.

The Human Cow posted:

I can't see that because I don't have a subscription, but I was pointed to this on Server Fault: https://wiki.centos.org/TipsAndTricks/CreateNewInitrd. Does that look like it's a good idea?

You can get a free developer account from Red Hat which should give you access to the docs. It would also enable you to run a bunch of virtual RHELs on your own computer, which might be useful in your work.

CaptainSarcastic
Jul 6, 2013



Here's another dumb question about my pending upgrade: how dumb would I be to use Btrfs instead of ext-4?

OpenSUSE defaults to wanting to use Btrfs, and I have always changed it to use ext-4 instead. The snapshot function with Btrfs is kind of tempting me, especially since I'm now planning on running Tumbleweed this time around. This is for desktop use, on a machine that shuts down and reboots at least once daily.

I'm pretty sure the answer is to use ext-4 and back-up using other means, but I figured I'd throw it out there anyway.

Mr Shiny Pants
Nov 12, 2012

CaptainSarcastic posted:

Here's another dumb question about my pending upgrade: how dumb would I be to use Btrfs instead of ext-4?

OpenSUSE defaults to wanting to use Btrfs, and I have always changed it to use ext-4 instead. The snapshot function with Btrfs is kind of tempting me, especially since I'm now planning on running Tumbleweed this time around. This is for desktop use, on a machine that shuts down and reboots at least once daily.

I'm pretty sure the answer is to use ext-4 and back-up using other means, but I figured I'd throw it out there anyway.

I'd rather use EXT4 than BTRFS to be honest. It never struck me as "stable". IMHO though.

BTW: I now have wireguard running with NAT on the server interface, so all traffic seems to be coming from a machine in the LAN instead of being routed. It works, even SMB, but I don't quite know if I like it or not.

Pretty nifty stuff.

Antigravitas
Dec 8, 2019

Die Rettung fuer die Landwirte:
Single disk btrfs is probably okay or they wouldn't default to it. If you take out all the bits that replace mdraid/lvm you are left with a snapshot-capable COW file system where you want to disable COW on many files. Transparent compression may be another feature that could be interesting. We have a good 1.2 compression ratio across all our files, so it could be an easy win. (Using ZFS and lz4, we don't use btrfs)

It is apparently reliable enough for single-disk usage, but that's a rather low bar…

BlankSystemDaemon
Mar 13, 2009



BTRFS, designed to compete with ZFS, and it's "reliable enough for one disk" :laffo:

taqueso
Mar 8, 2004


:911:
:wookie: :thermidor: :wookie:
:dehumanize:

:pirate::hf::tinfoil:
Imagine a world where Hans didn't murder anyone

waffle iron
Jan 16, 2004
I'm excited for bcachefs getting into mainline kernels but that will be some time away. Until then ext4 is what I'm sticking to.

xtal
Jan 9, 2011

by Fluffdaddy
I've used btrfs for an extremely long time, even back when it was labeled as unstable, and haven't had any problems with it. Posting this from btrfs. I chose it for backups and COW, but subvolumes have hidden benefits (see how Docker uses them) and the transparent compression is nice.

ZFS is probably better, especially for complicated workloads. But contrary to all of that, on fresh installs I use ext4, since you can manage them just with UNIX tools and not filesystem-specific programs like zfs and btrfs.

Not Wolverine
Jul 1, 2007
I decided to give openSUSE tumbleweed with KDE a try in a VirtualBox VM, the first thing I've noticed is that the Discover software center claims it is lacking network connectivity. Upon searching google, the first result I found mentions packagekitd but also metions that this might interfere with zypper. I don't know enough about YaST to have an opinion of whether or not it's good, but I just like the GUI for Discover a little better for seeing what's available, like I don't yet know if I want a GTK or QT torrent client, or if I might ever like another PDF program better than Okular.

Should I be able to use the Discover software center without causing other issues with the system? I'm more familiar with Ubuntu, I just want to try SUSE a little since I think it's a more professional or commercial Linux, kinda like Fedora and Redhat. Would I notice much difference between openSUSE and SUSE Enterprise Linux? Is there any sort of free desktop download for SUSE Enterprise Linux besides openSUSE? I might be wrong, but legally, can a person install RHEL on a desktop computer, and use it, but just not have any sort of tech support? I am under the impression the primary difference between "enterprise" and desktop Linux is that enterprise is more stable due to having significantly older, more tested packages that might not support the latest features, is this accurate?

Antigravitas
Dec 8, 2019

Die Rettung fuer die Landwirte:
If you want a personal RHEL you use CentOS, which is RHEL with the branding replaced.

I'd assume SuSe does something similar but I'm not familiar with their stuff.

Discover is garbage imho, for the same reason the new Windows Update GUI is garbage. It sits upon a layer of indirection and simply cannot cope or offer useful feedback when things don't work.

Also, what you call KDE is actually Plasma from the KDE people, or as I like to call it, KDE/Plasma… :v:
And you want the QT stuff over GTK, because QT is what Plasma and the KDE ecosystem uses. GTK is a foreigner in that system and looks weird no matter how much effort the KDE people put into theming it.

CaptainSarcastic
Jul 6, 2013



Discover looks neat in theory but I avoid using it entirely.

It has a propensity to appear hung, but in reality it just takes a year and a day to do anything. Using Yast and the software manager is a much better approach, or pulling in stuff from https://software.opensuse.org/explore if it isn't already in one of the established repos.

Edit: Oh, and I did install Tumbleweed on my new machine, and did stick with ext4. It's running beautifully so far - no complaints.

Worf
Sep 12, 2017

If only Seth would love me like I love him!

I checked the OP for this,

I heard you can make a bootable linux USB with persistent storage. that would be useful for me. is there a webpage i can use for a step by step on this? ive tried it a couple times and dont seem to have the right results

i'm trying to use ubuntu since im familiar with it but i guess if i was forced into something else that would be ok

Shaocaholica
Oct 29, 2002

Fig. 5E

waffle iron posted:

I'm excited for bcachefs getting into mainline kernels but that will be some time away. Until then ext4 is what I'm sticking to.

Cool, patreon file system. I can support that!

e: how CPU intensive is bcachefs compared to ext4? On a simple single disk config.

e2: oh

https://www.phoronix.com/scan.php?page=article&item=bcachefs-linux-2019

Shaocaholica fucked around with this message at 18:04 on Apr 22, 2020

Antigravitas
Dec 8, 2019

Die Rettung fuer die Landwirte:
Phoronix's fs tests are typically hot garbage though.


Statutory Ape posted:

I heard you can make a bootable linux USB with persistent storage.

I remember you could do this but it may have been deprecated because it was kind of hacky.

However, nothing stops you from just installing the OS on a USB drive. It would behave just like any other installation, though I'd use a more expensive drive for this since the flash chips in thumb drives are typically of much lower endurance.

Mr. Crow
May 22, 2008

Snap City mayor for life

Antigravitas posted:

Phoronix's fs tests are typically hot garbage though.

How do you figure? It's using FIO which is THE standard is it not?

Worf
Sep 12, 2017

If only Seth would love me like I love him!

Antigravitas posted:

Phoronix's fs tests are typically hot garbage though.


I remember you could do this but it may have been deprecated because it was kind of hacky.

However, nothing stops you from just installing the OS on a USB drive. It would behave just like any other installation, though I'd use a more expensive drive for this since the flash chips in thumb drives are typically of much lower endurance.

So i kind of accidentally did this. I tried to load the OS from one PC to another using an NVME m2 to tb3 adapter. now i can move that external from system to system and it works like i want, but its too cumbersome for the application i need rn.

how tf did i do that///can i replicate it onto a flash drive?

did i just reboot and then select that drive / partition to install it on... and its that easy?

i assume more people cant/dont do that because hardware differences from install to install would cause issues? thankfully all the PCs in question are running 8th or 9th gen I7s with intel wifi, except one HP i have that also works just fine

Computer viking
May 30, 2011
Now with less breakage.

The easiest is probably to use two usb sticks - boot from one, install to the other. That way it can just do the full "set up an empty drive however it wants"-routine. You may want to pull any other drives in the system first, just to make sure it doesn't try to be smart with installing grub on what it assumes is the boot HD.

BlankSystemDaemon
Mar 13, 2009



Mr. Crow posted:

How do you figure? It's using FIO which is THE standard is it not?
It's hot garbage because Michael does a single run, doesn't use ministat so can't report on minimum, maximum, mean, or median - nor does he apparently know what statistics confidence or Student Ts are, since his graphs don't even feature a margin of error.

So basic stuff that ought to be taught in school, nothing special.

Antigravitas
Dec 8, 2019

Die Rettung fuer die Landwirte:

Mr. Crow posted:

How do you figure? It's using FIO which is THE standard is it not?

There are tons of options for fio. Just using some program doesn't make a good benchmark.

The first benchmark is already highly dubious. Application startup measuring the start of loving XTERM?! And right there you already see ZFS miraculously outperforming everything to a degree that should immediately halt all testing until he figures out where that anomaly comes from. (It's probably because the cache is hot because if he's dropping caches at all, he's only dropping the page cache and not ZFS's ARC)

His benchmarks are full of those unexplained anomalies. I can almost guarantee he's using 512e on that Optane when literally every single sane storage admin will use at least 4ke and configure ZFS for 4k blocks. Benchmarking this difference would be interesting, but Phoronix doesn't do that. But who knows, it's not documented so the benchmark can't even be replicated…

BlankSystemDaemon
Mar 13, 2009



Rebooting between benchmarks? It's more useful than you think!
Unless you're testing the hot cache performance, of course.

Mind you, it makes very little sense to test any CoW tree-based filesystem against traditional filesystems based ultimately on the Fast File System from BSD ie. EXT*.

Computer viking
May 30, 2011
Now with less breakage.

D. Ebdrup posted:

It's hot garbage because Michael does a single run, doesn't use ministat so can't report on minimum, maximum, mean, or median - nor does he apparently know what statistics confidence or Student Ts are, since his graphs don't even feature a margin of error.

So basic stuff that ought to be taught in school, nothing special.

PTS defaults to three runs, I believe, and do more if the results differ by some percentage. If you can find his benchmarks in the openbenchmarking viewer, it reports mean, N and stddev.

Which raises the question of why he has decided to hide this in the article form (except for the error bars he puts in for results with large variance) - I believe it's more "dumbing down to a general audience". Which would be less of an issue if he was more consistent about linking to openbenchmarking in every article.

And in the context of "you get to choose between these file systems on your workstation" it makes a lot of sense to compare the alternatives with different workloads. If both are alternatives I can realistically pick and neither have any big disqualifying marks then ... apples and oranges are both fruit, and can absolutely be compared.

Computer viking fucked around with this message at 18:49 on Apr 23, 2020

capitalcomma
Sep 9, 2001

A grim bloody fable, with an unhappy bloody end.
Well poo poo, Ubuntu 20.04 comes with a new version of unison which isn't backwards-compatible with the previous build (that my other PC uses).

What's involved in running newer packages than what's in the repo in Debian stable? Or running older snaps on Ubuntu? Is it worth the effort, or should I just put 20.04 on the other PC as well?

capitalcomma fucked around with this message at 03:53 on Apr 27, 2020

Adbot
ADBOT LOVES YOU

BlankSystemDaemon
Mar 13, 2009



POLA, in Your Unix? It's less likely than you think!

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply