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
Tad Naff
Jul 8, 2004

I told you you'd be sorry buying an emoticon, but no, you were hung over. Well look at you now. It's not catching on at all!



... Oregano? Is this some amazing utility I've never been exposed to?

Tad Naff fucked around with this message at 18:46 on Feb 20, 2021

Adbot
ADBOT LOVES YOU

VictualSquid
Feb 29, 2012

Gently enveloping the target with indiscriminate love.


every true unix wizard has a whole bag of dates in easy reach.

Antigravitas
Dec 8, 2019

Outside Context Problem


Well, if you read comp.unix.wizards, as we all do, you'd know this:

quote:

The oregano is reputedly referring to an incident in which one of the
original folks involved with BSD was hassled for coming across the
Canadian/U.S. border with a bag of what was assumed to be an illegal
substance, and turned out to be oregano.

https://groups.google.com/g/comp.unix.wizards/c/qkiqSJWgEPE/m/E5BwZYMvXwAJ

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



I sent this.

Antigravitas
Dec 8, 2019

Outside Context Problem


Nice. And I found a fairly high res scan of that poster floating around. Let's see if I can get this printed and shipped to me.

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Antigravitas posted:

Nice. And I found a fairly high res scan of that poster floating around. Let's see if I can get this printed and shipped to me.
The one linked wasn't high-res enough?

Antigravitas
Dec 8, 2019

Outside Context Problem


The one I found was a gigantic png that may be the source of the jpg. Or it may be someone who turned the jpg into a png because png better.

But no, need MOAR res, I wanna blow it up.

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Antigravitas posted:

The one I found was a gigantic png that may be the source of the jpg. Or it may be someone who turned the jpg into a png because png better.

But no, need MOAR res, I wanna blow it up.
Can you link the png?

Mr. Crow
May 22, 2008

Snap City mayor for life


BlankSystemDaemon posted:

Can you link the png?

.


Also I was gonna comment on the posters with weird desktop behavior, try making sure you're not on Wayland. Despite the devs insisting otherwise it's not ready for prime time yet, not for most users.

Newf
Feb 14, 2006
I appreciate hacky sack on a much deeper level than you.

Can anyone offer comment on how I could expect an Ubuntu install to do on either:

- Lenovo X1 Carbon (7th gen), 14''
- Lenovo X1 Yoga (5th gen), 14''

And anyone with current hardware knowledge can also feel free to compare these machines to the MacBook Pro 13 and 16 models.

I've not owned a Mac before, but I think that the MBPs here may be superior machines out of this bunch, and I'd be silly to take the PC option.

?

Antigravitas
Dec 8, 2019

Outside Context Problem


It's this one: https://archive.org/details/unix-magic-poster-gary-overcare-1

It doesn't have compression artefacts around the text so I assume it's the source of the jpg (same res). It's still a bit blurry so I'll have to do some cleaning up when I blow it up to proper DPI (I never trust printing shops to do this right).

I'm trying to find the exact font they used, but I have a hunch that at least the UNITECH logo comprises of glyphs not found in a normal font.

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Antigravitas posted:

It's this one: https://archive.org/details/unix-magic-poster-gary-overcare-1

It doesn't have compression artefacts around the text so I assume it's the source of the jpg (same res). It's still a bit blurry so I'll have to do some cleaning up when I blow it up to proper DPI (I never trust printing shops to do this right).

I'm trying to find the exact font they used, but I have a hunch that at least the UNITECH logo comprises of glyphs not found in a normal font.
I found it elsewhere, and tried to embiggen it, but yours sounds much more involved, so I'll be very interested in what you end up with.

Antigravitas
Dec 8, 2019

Outside Context Problem


It's now on my list of things I want to get done, but it definitely jumped the queue.

Also part of that list:

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Antigravitas posted:

It's now on my list of things I want to get done, but it definitely jumped the queue.

Also part of that list:

is right.

CaptainSarcastic
Jul 6, 2013

HAIL SATAN


BlankSystemDaemon posted:

You forgot the two others:



I really like the first of these, but the second one makes me want to slap his hand.

Don't touch my monitor screen. Ever.

I've always hated people touching display screens going back to CRT days, and with LCDs it's even worse. I actually do okay with touchscreens, like on my Chromebook, but that is a very specific exclusion. Otherwise it is a very simple and absolute rule.

The Milkman
Jun 22, 2003

No one here is alone,
satellites in every home


Antigravitas posted:

It's this one: https://archive.org/details/unix-magic-poster-gary-overcare-1

It doesn't have compression artefacts around the text so I assume it's the source of the jpg (same res). It's still a bit blurry so I'll have to do some cleaning up when I blow it up to proper DPI (I never trust printing shops to do this right).

I'm trying to find the exact font they used, but I have a hunch that at least the UNITECH logo comprises of glyphs not found in a normal font.

Oh yeah I think that's where I got it from. I had to jpg it to upload to imgur. I also assumed people did not want to load a 35mb png file

Antigravitas
Dec 8, 2019

Outside Context Problem


The Milkman posted:

Oh yeah I think that's where I got it from. I had to jpg it to upload to imgur. I also assumed people did not want to load a 35mb png file

You assumed wrong, bucko.

Newf posted:

Can anyone offer comment on how I could expect an Ubuntu install to do on either:

- Lenovo X1 Carbon (7th gen), 14''
- Lenovo X1 Yoga (5th gen), 14''

And anyone with current hardware knowledge can also feel free to compare these machines to the MacBook Pro 13 and 16 models.

I've not owned a Mac before, but I think that the MBPs here may be superior machines out of this bunch, and I'd be silly to take the PC option.

?

In general: Intel / AMD integrated graphics and Intel Wifi give you the least troubles. Lenovo, afaik, tries to use components from vendors that support Linux, so perhaps even the vendor-garbage works.

I don't quite get why that carbon is so expensive with its woefully underspecced RAM, but then I generally don't really understand why you'd buy an expensive laptop anyway.

Newf
Feb 14, 2006
I appreciate hacky sack on a much deeper level than you.

Antigravitas posted:

In general: Intel / AMD integrated graphics and Intel Wifi give you the least troubles. Lenovo, afaik, tries to use components from vendors that support Linux, so perhaps even the vendor-garbage works.

I don't quite get why that carbon is so expensive with its woefully underspecced RAM, but then I generally don't really understand why you'd buy an expensive laptop anyway.

Hmm. Thank you. These are options provided to me at a new job.

Newf fucked around with this message at 13:07 on Feb 21, 2021

Vavrek
Mar 2, 2013

I like your style hombre, but this is no laughing matter. Assault on a police officer. Theft of police property. Illegal possession of a firearm. FIVE counts of attempted murder. That comes to... 29 dollars and 40 cents. Cash, cheque, or credit card?

Is there a way to implement Microsoft-style scrollbar snap behavior? (Running Arch, KDE.) The more I think about this, the more I suspect the answer is "no", but it's better to ask.

If you're unfamiliar with what I mean, it's well described here by someone who hates it:

quote:

Now, moving on to the actual problem at hand. The problem I see is that during scroll blob drag operations, Windows creates an invisible “hotspot” around the scrollbar; if you move your mouse cursor outside of this hotspot while still dragging the blob, it will reset itself to its initial position. If you’re still holding down the mouse button, the reset is visual only; moving your mouse pointer back into the hotspot will “undo” the reset. If you release, however, the reset will become permanent.

And discussed here as well.

I use this all the time as a way of putting my finger down on one spot in a long webpage, flipping back and forth to check something (byline of an article, how far up in the page comments on that article start, for instance), and returning to my current spot.

I've found accounts that KDE adopted this behavior back in 1999, but comments on the first article from 2009 say KDE 3.5 didn't behave that way. I guess it was a fad.

Computer viking
May 30, 2011
Now with less breakage.

The Carbon is expensive because it's a thin, very light, solid laptop with a good keyboard, and good enough specs. I went from an older Carbon to a T14, and while I appreciate having a more powerful and upgradeable machine for a better price, I miss it every time I walk or bike home from work.

Computer viking
May 30, 2011
Now with less breakage.

Speaking of Thinkpads: I just did a silly thing, and the recovery method may perhaps one day be useful to someone.

I tripleboot this T14 windows/FreeBSD/Windows. The FreeBSD install is not very happy right now; the DRM drivers for the Vega 8 GPU is not quite working yet.

The FreeBSD install is root-on-ZFS, of course. It also has a goodly amount of free space, while the Fedora install is a bit cramped. I thought I'd import the zpool and create a dataset for a few steam games - so I did
zpool import -a
zfs create zroot/zfs_steam
zfs set mountpoint=/zfs_steam zroot/zfs_steam


So far so good, until everything stopped working about two seconds later. Apparently, mounting the FreeBSD versions of /usr and /bin and everything on a running linux does not work very well. Adding insult to injury, ZoL is actually competent enough to automatically mount ZFS pools on boot, if you have imported them once (or created them on the same Linux install, ofc). Which means it persists on reboot.

Oh, and I don't have a root account on Fedora, it's all modern and sudo-based - which means the rescue boot in grub doesn't get me a working shell. However, editing a normal boot to add init=/bin/bash to the kernel still gets you something. From there, you can
mount -o remount,rw /
rm /etc/zfs/zpool.cache
sync

From there, the smoothest way to reboot is ^D or "exit", which kills bash and elegantly segfaults the kernel, leaving you in a state where your only option is to hold the power button until the laptop turns off.

On reboot, remember to use -R /zfs on the zpool import, to set an altroot that it will mount everything under.
(This does not create a cache file, and does not automatically re-import at boot. I'm not sure what the best way to get both that and an altroot is.)

edit: Also, don't forget that Steam is a goddamned flatpak and you need to sudo flatpak override --persist com.valvesoftware.Steam --filesystem /zfs/zfs_steam or something to that effect, or Steam will appear to work but mysteriously not see the zfs mountpoint.

At this point this is just notes for myself in the future. All I wanted was to continue my Darkest Dungeon save on the laptop even though the sound is broken again in Windows. The good news is that Fedora running Steam-for-Ubuntu in a flatpack with the game data on a FreeBSD zpool mounted with ZoL ... just works, when you get there. Perfectly, even.

Computer viking fucked around with this message at 01:04 on Feb 23, 2021

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Computer viking posted:

zpool import -a
You're a brave person for not using -R, which is made to change the root folder of where other datasets are mounted in relation to.

Computer viking
May 30, 2011
Now with less breakage.

BlankSystemDaemon posted:

You're a brave person for not using -R, which is made to change the root folder of where other datasets are mounted in relation to.

Brave, in that very specific sense of "tired enough to not worry about consequences, yet awake enough to still be dangerous" sense, yes.
And yeah, I did edit in a line about -R a bit later in the evening.

BlankSystemDaemon
Mar 13, 2009

System Access Node Not Found



Computer viking posted:

Brave, in that very specific sense of "tired enough to not worry about consequences, yet awake enough to still be dangerous" sense, yes.
And yeah, I did edit in a line about -R a bit later in the evening.
To paraphrase Heinlein, it's surprising how much getting older resembles being too tired - and yeah, I definitely know the feeling.

Also, freebsd has parallel service handling now (in addition to parallel jails, which it's had for 8 years on my next birthday.
It won't be in 13.0-RELEASE, but it'll likely be in 12.3-RELEASE and 13.1-RELEASE.

Computer viking
May 30, 2011
Now with less breakage.

Oh, that's very neat - I've vaguely looked at how the rc scripts work before, and got to "Ok, so we can build a dependency tree from this, sort of like what Make does - so in theory we can do things in parallel without really rewriting the scripts (much)". It did not seem like a project I wanted to start on, though.

(It'd be "fun" to create a system that booted using straight up Makefiles, though - at least you'd get the parallelization for free.)

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!



Do any of you guys work in HPC? Thoughts?

Computer viking
May 30, 2011
Now with less breakage.

Bob Morales posted:

Do any of you guys work in HPC? Thoughts?

No, I'm just below that in hardware requirements, though we sporadically touch those systems. Still - what sort of things do you want to know? It's a very open-ended question.

rufius
Feb 27, 2011

Clear alcohols are for rich women on diets.


Bob Morales posted:

Do any of you guys work in HPC? Thoughts?

A very long time ago. But it was academic and, weirdly, I was managing and writing software for an Apple Xserve cluster.

Did you have a specific question or...?

Bob Morales
Aug 18, 2006


Just wear the fucking mask, Bob

I don't care how many people I probably infected with COVID-19 while refusing to wear a mask, my comfort is far more important than the health and safety of everyone around me!



Computer viking posted:

No, I'm just below that in hardware requirements, though we sporadically touch those systems. Still - what sort of things do you want to know? It's a very open-ended question.

rufius posted:

A very long time ago. But it was academic and, weirdly, I was managing and writing software for an Apple Xserve cluster.

Did you have a specific question or...?

Just kind of wondering what types of things you guys do, what the biggest differences are from traditional client/server issues, not really sure I guess. I have an interview coming up for an HPC position and just want to know a little more about it.

Computer viking
May 30, 2011
Now with less breakage.

Right, that way. My impression looking in is that the backend people are mostly worried about rack infrastructure, bandwidth, the wildly expensive enterprise storage systems, and how to provide network bootable images with new patches but no service interruption, while the more user-facing people get to enjoy queueing systems and whatever tools it is that lets a user say "I need python3 with numpy for this job" and have it automatically imported into the environment - and the joy of packaging whatever new versions and software the users want to run.

Which is probably not super surprising.

rufius
Feb 27, 2011

Clear alcohols are for rich women on diets.


Bob Morales posted:

Just kind of wondering what types of things you guys do, what the biggest differences are from traditional client/server issues, not really sure I guess. I have an interview coming up for an HPC position and just want to know a little more about it.

So, major caveat is this was probably 13+ years ago.

Things that were different at the time:

  • Apple Xserve still existed
  • virtualization wasn't yet a widespread thing
  • Tools like Salt/Chef/Puppet didn't really exist in any wide-scale popularized fashion. Or at least I don't remember it.

So with that context, I'd say a lot of the problems were "I need to make sure all 24 of these Xserves have the same configuration." Apple gave some tools for the mainline workflows for configuring that, but anything specialized was a roll your own story. I ended up homebrewing a script locally in python that would copy over a Python script to each server and use it as the orchestrator for various commands/edits. I don't think that's wildly different than larger scale configuration of racks of servers then or today.

The biggest thing I recall is spending a lot of time making sure the server-to-server interconnects were well configured. I don't remember what we were using - Infiniband maybe? But no one knew anything about it and so I was a 21 year old trying to figure out wtf to do with it all.

TL;DR - It's a lot of similar sorts of tasks as typical server administration. The biggest difference, is that at a machine layer instead of an application layer, they're interacting more. So you might have more weird scenarios to deal with there.

xzzy
Mar 5, 2009



I'm a "fringe admin" (meaning I have root and can help out, but am far from an expert on doing anything with it) of a moderate sized HPC cluster and they still rely heavily on infiniband, even though it appears to be a dead end tech. You want cables? Prepare to wait six months for them to be custom made and shipped from China! The transfer rates are fuckin' bonkers though, so that's nice.

It's fine I guess, in general it's not any different than any other batch cluster though the users tend to want more cutting edge software and are extremely sensitive towards anything using cpu cycles that isn't their job.

Pablo Bluth
Sep 7, 2007

I've made a huge mistake.


Bob Morales posted:

Just kind of wondering what types of things you guys do, what the biggest differences are from traditional client/server issues, not really sure I guess. I have an interview coming up for an HPC position and just want to know a little more about it.
I have a Computational Fluid Dynamics background.

Interconnects between the nodes is really important. Hardware-wise that means Infiniband for high bandwidth and low latency. Softwarewise; a lot of software relies on MPI for the internode communication.

Queueing systems/schedulers for allocating resources to users; PBS/TORQUE or Slurm.

How to deal with users filling up all the disk storage or queuing jobs that take resources they're not mean to.

It's slightly out of date but this works as an overview of what a HPC system is trying to deliver. Or at least for CFD.
https://storage.ansys.com/corp/2012/April/it/it_guide.pdf

HPC Wire
https://www.hpcwire.com/

Oh, and the systems will invariable break at 6pm on Friday when the users want to have their stuff running over the weekend. It won't ever break in the morning...

Pablo Bluth fucked around with this message at 20:32 on Feb 23, 2021

Methanar
Sep 26, 2013
ASK ME ABOUT NOT TIPPING DELIVERY DRIVERS, OR ABOUT MY DIET OF CANNED BABY CORN AND CHICKEN NUGGETS

What is the difference between HPC and just running a bunch of "normal" servers each hosting microservices talking to each other over a message queue?

Also reading this C example of MPI, this at the surface level to me looks similar to golang channels for passing messages around between goroutines. Is that a fair comparison as to how it works and what it may be used for?
https://en.wikipedia.org/wiki/Message_Passing_Interface

Methanar fucked around with this message at 20:38 on Feb 23, 2021

Antigravitas
Dec 8, 2019

Outside Context Problem


Your typical HPC workload is characterised by a requirement for low-latency and high-bandwidth inter-node communication.

Weather simulation is a good example. A naive approach cuts the earth into smaller cells so that the working set fits into the worker node's RAM. However, each node will have to exchange information with each of its neighbours after each step of the simulation before they can continue, since neighbouring cells affect each other. That chatter can be significant and is highly latency-sensitive. If it takes ages (like, say, 10ms) your simulation slows to a crawl.

That's one thing you can't really do on commodity hardware. If you look at what people do on HPC you'll find tons of simulation workloads because they all have this problem.

Pablo Bluth
Sep 7, 2007

I've made a huge mistake.


Methanar posted:

What is the difference between HPC and just running a bunch of "normal" servers each hosting microservices talking to each other over a message queue?
Scale.

edit: another resource is the HPC Advisory Council's published conference talks and best practice guides.
https://www.hpcadvisorycouncil.com/conference_publications.php?pg=published_works&sub=conferences

Pablo Bluth fucked around with this message at 21:16 on Feb 23, 2021

Computer viking
May 30, 2011
Now with less breakage.


Not only - like Antigravitas says, it's as much about the fast/wide interconnects. BOINC and Folding@home have massive numbers of nodes, but I don't think they are typically counted as HPC clusters?

xzzy
Mar 5, 2009



Pretty much every government HPC cluster was given a mandate to start running protein folding jobs in the last year too. I bet you can't guess why!

Pablo Bluth
Sep 7, 2007

I've made a huge mistake.


F@H has the 'large number of nodes' element of scale but the communication is limited to sporadic communications between the home user and the F@H service. My point was that HPC workloads tend to meet all the elements of scale; lots of nodes talking verbosely nonstop for a long time.

Pablo Bluth fucked around with this message at 21:30 on Feb 23, 2021

Adbot
ADBOT LOVES YOU

Methanar
Sep 26, 2013
ASK ME ABOUT NOT TIPPING DELIVERY DRIVERS, OR ABOUT MY DIET OF CANNED BABY CORN AND CHICKEN NUGGETS

xzzy posted:

Pretty much every government HPC cluster was given a mandate to start running protein folding jobs in the last year too. I bet you can't guess why!

realized fears of vulnerability to bio terrorism?

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