LINUX Unplugged - 363: Return of the Terminal Server
Episode Date: July 22, 2020Fedora makes a bold move and Microsoft seems to be working on their ideal "Cloud PC", we ponder what Linux has to offer. Plus an easy way to remotely watch movies with others, and a bunch of your feed...back. Chapters: 0:00 Pre-Show 0:29 Intro 2:34 RSI Woes 7:21 Jellyfin 13:12 Proxmox Backup Server 16:22 ProcMon for Linux 20:16 Fedora 33 Defaults to Btrfs 24:27 systemd-oomd 26:38 Housekeeping 28:41 Riot Becomes Element 32:36 Mysterious Microsoft Job Posting 39:19 Picks: Polybar 40:57 Picks: ytop 42:43 Feedback 48:50 Outro 50:06 Post-Show Special Guests: Brent Gervais, Drew DeVore, and Neal Gompa.
Transcript
Discussion (0)
shop.gnome.org.
Should I say gnome or gnome?
You know, I was wondering that.
When you're referring to a web URL,
do you say...
Which one do you say?
Gnome or gunome?
I would say gnome, but it's just me.
Who cares? They have a merch store and that's all that matters.
Hello friends and welcome into your weekly Linux talk show.
My name is Chris.
My name is Wes.
Hi Wes. Wow. Did you put those fancy boots on for me?
Of course I did. I wanted to welcome you back from your grand adventure. Thank you. It is really great to be back and we have a great show. So much to catch up on. Some solid feedback and some great
picks this week. It's a real classic. And before we get any further, I want to say hello to Drew
and Cheese. Hello, gentlemen. Thanks for being here. Well, hello there. Hello. Hey, you both
are also quite dressed up. Cheese, I don't think I've ever seen you in a suit before. And a bow
tie. I know, dude.
It's weird.
Yeah, I actually think the bow tie is working.
I think it's working.
Drew in a tie, though.
That's going to take some getting used to.
But you're supposed to put it around your neck, not your forehead, Drew.
Also, before we go, I have to say a huge hello, a hearty time-appropriate greetings to our
mumble room.
Hello, virtual lug.
Hello.
Hello.
Hello.
Happy Linux Tuesday. Happy Linux Tuesday to y'all. Hello, virtual lug. Hello. Happy Linux Tuesday.
Happy Linux Tuesday to y'all. And it's true, I'm back. Back in the Pacific Northwest. It's really nice to be back in the studio. traveling. Tricky keeping everyone healthy and safe during all of that while also visiting family
and working and, of course, trying to also make it a memorable experience for the kids and do the
shows. It was a lot, but it was maybe one of the greatest road trips of our family's lifetime. So
I really enjoyed it. And awesomely, while I was out and about, Angela managed to get the roof replaced on the studio.
So we got some really, really important work done for the studio that we've been needing to do for a long time,
taken care of while I was out for so long.
So that's pretty great.
Big old thanks to Ange for pulling that off because I never would have got it done.
No.
Or you would have scheduled it for the same day we were doing a Linux Unplugged.
So there would have been roofing noises in the recording.
Yeah, that was really the strategy of it.
But now that I'm back, I don't know, I guess I was so stupid.
I thought I would, I don't know, I would spend 50 days on the road and I would be in such different
conditions that when I got back, I wouldn't have RSI. But within a couple of hours of sitting back
at my desk, I was having significant bad RSI again. I mean, seriously, from my fingertips
into my wrists, into my elbows, all the way up to my shoulders. There's intense pain and I have trouble now holding stuff. So I need some RSI tips from the audience
because I was trying to pretend like this was going to go away if I just changed things up for
a bit. And it's like, even now I haven't really done, I've tried not to be at my desk a ton this
morning and my hands are stiff and sore. Exercises, anything people could give, because this is
starting to freak me out, because I'm worried I'm not going to be able to use the computer for more
than an hour at a time without hurting myself. So, I mean, I've had issues, and I know that
previously I was really bad about putting my elbow on my desk and then resting my chin on my elbow.
Yeah, I do that on my armchair.
resting my chin on my elbow. Yeah, I do that on my armchair. That's really bad. But one suggestion I would say is get yourself a motorized desk where you can stand and then alternate between
sitting and standing. I do that about every hour, hour and a half. Sit down for an hour and a half,
stand up for an hour and a half. And that seems to help out a lot. You cannot buy any personal
fitness stuff on Amazon really right now because of the pandemic, but you can pick up a, it's like an
elastic band set that you just throw over the door. And then that way you can do like a couple
of, you know, like different kinds of strokes using these bands. And it'll help out a lot too.
It'll strengthen all the muscles that are kind of in the front of
your chest and down your arms and your biceps and help out a little bit. So that's helped me out a
lot, but you know, ultimately I did have to have a surgery related to my neck and I don't know if
it's connected to that or not years of being at a computer desk, but certainly could be.
Neil, some of us Linux users are just aging a bit. I think it's going to be something
that happens to a lot of us.
Is this something you've struggled with?
I've dealt with this quite a lot.
It got much worse once I started working from home
because my setup at home is not optimized
for continuously doing the same kinds of things I do at work.
It's actually more optimized for me
to just sit and play games.
Right. Yeah. Chill.
My desk is actually a little too tall. So typing is like quite painful. So a couple of things that
I wound up doing. The first was I did some awful work to slightly lower my desk a little bit so
that the height wasn't exactly at eye level for me for the – because my – I have no stands or auxiliary whatevers.
But like getting my keyboard to be lower so that I don't always have to like prop up my arms to type helped.
Also getting a very soft wrist rest across the whole keyboard so that whenever I'm typing, it can rest and not feel like I'm digging into my wrists.
Having an ergonomic keyboard, making sure you have a good wrist rest,
actually making it so that you don't have to prop up your arms
to like type on your keyboard also makes it so that you have less,
you're less likely to like dig into and put pressure across everything,
at least in my experience, that has made it better.
Again, not medical professional, please go seek a doctor. But so that's helped for me.
It's kind of fascinating. It just makes me think about when I was a youngster taking piano lessons
and you know, you got to learn the proper form. And despite many of us being professionals or
just hobbyists, either way, spending all of our time sitting at the computer, we don't think that
we need to invest in that and, you know, get the forms right if we're going to be doing it all the time i wish i had
taken it seriously sooner because it's gotten to the point where uh i was out doing some yard work
in the junkyard just because while i was gone it turned into a jungle and even 15 minutes of holding
uh the whacker thingy my my elbow was done for the night.
It was really, it just stinks.
So I got to get ahead of it.
Or at least, I mean, maybe I'm not ahead of it anymore, but I got to do something about it.
Anyways, I just wanted to put it out there because I think a lot of us who are into Linux are very intense about the amount of time we spend at the keyboard.
are very intense about the amount of time we spend at the keyboard. And I do want to try to get the word out there because I think one of these, it's one of these things, if you don't
take care of it before it starts, once it starts, you're at basically a stage of mitigating and
managing it from that point forward instead of preventing or curing it. Maybe I'm wrong,
but that seems to be what I'm discovering. Moving on, though, because we have some really good news.
And that's Jellyfin has a brand new release that looks pretty great in general. But one thing that Wes and I both noticed is it introduces a new feature called Sync Play.
Sync Play allows you to create a room that other users can join in order to share essentially a common viewing experience.
And there seems to be no limit on the number of users in a room. You're all free to join. It's Jellyfin. So there's a
login, but you can share that same user account or you could be the same user on multiple devices.
And we had a chance to try it out this morning. Yeah, we did. I mean, this is something that both
you and I have been looking for. Now there there's various solutions for, there's a plug-in system for VLC, there's stuff for Netflix, there's stuff for YouTube, but none of them really kind of hit the sweet spot. I think Jellyfin Sync Play is getting close because you and I were with Jellyfin, it's a media management system for
your videos. It indexes them. It goes out and gets the metadata. A lot like Plex. One thing I like is
it's, you know, it's just a little more lightweight. Plex is amazing. Don't get me wrong. But, you know,
it's tied into the company. It's tied into the web servers. Jellyfin was forked off MB. And, you know,
you just start up the Docker container, get it going. You can make a guest account. You can make
local user accounts. So it's really easy to share with people, even if they're not totally bought into the whole ecosystem.
And now the support's getting way better.
It works with the Chromecast.
I don't think it has integration with many of the clients yet, but hopefully that will come down the road. And really a huge thank you both from me, from Chris, and I think from the Jellyfin community to first-time contributor Andre Hoanka, who's a computer science student over at a university in Italy.
He submitted pull requests for both the server changes and the web client changes that laid the foundations for all of this.
And I think it's a really cool open source success story because there's actually been like a feature request for something like this feature since all the way back till like 2018.
And it just took someone stepping up, doing the work, putting it in.
And then the project was, you know, happy to include it and actually shipped it out where we can all go test it and abuse it.
Yeah, and the timing's pretty good, right?
Because this is a time when people maybe would get together and watch something that are now spending time apart, but there's still options like this to do some kind of remote
viewing party. And the great thing is, is there wasn't something that was free software and readily
available and didn't require cloud services until this came along. So I'm really glad the Jellyfin
project proceeded with implementing this
because I know there was some people in the community thought, oh, maybe this isn't something
we should really do. We don't think we can get it right. But I'm here to tell you it works pretty
well. Wes and I were pretty quick to get it set up and going. We both had control over the player.
We could skip ahead and skip back, play, pause, and it worked really well. It seems like one of
the little tricks they do is when you skip ahead, they pause the playback for everyone, which probably gives the players time to sync up.
And then when you hit play, by then they've already kind of figured out their positioning and it picks right back up.
When you consider how far apart folks might be, the various latencies introduced between locations and how close you are to the actual server, it's pretty impressive.
There were a few little details that you kind of have to get used to.
And I did find in my own testing that, you know, if someone's watching something and
you join in, the sync will be a little rougher.
But once you've paused and then resumed playback, it was really tight.
Yeah.
And we immediately just started chatting in a separate text chat about the show we were
watching.
There's no chat aspect to it.
So you provide that separately.
And I like that.
I don't want them to try to integrate some sort of WebRTC voice chat into this thing.
Just focus on the video playback, and we'll use the existing communications channel we
already had to set the whole thing up in the first place.
There's also just traditional improvements.
They're working on a more modern web client.
It's had a lot of technical debt.
It's really kind of just coming from a fork of MB and then implementing it themselves.
There's just old technologies in there at this point.
So some of those until recently prevented them from being able to use modern JavaScript tooling,
which they have taken a whacking to, thanks to a contributor who worked on improving
the way they build the web client.
They can now get to some of this.
Also, Wes, it looks like some of this
greatly simplifies the support
for some of their legacy clients.
And the reason why that's sort of notable
is that means they could extend support
back to webOS and Tizen versions of the client.
And I just love that
because if you've got devices
like that laying around and they have the means to make it work, I think it's pretty great to
see them put the effort into it. Yeah, you know, once you've got some of this additional tooling
available, that means you can target older versions of JavaScript, you know, other ancient
clients that they still have the hardware to do it. They just don't support the latest standards.
But with a tooling cleanup, well, that can be a lot better. And actually, you know, this is a pretty big change. Once it's
complete, it should make the web client a lot simpler. I already think the web client's, I mean,
pretty decent, although you can tell there's some rough edges. Sounds like eventually Jellyfin
wants to migrate over to Vue, one of the hot new JavaScript frameworks, and really just, you know,
do a good summer cleaning
of all this code that they've inherited. Lots of good stuff, lots of old stuff, but if you want it
to be shiny and fast and new, you got to make sure you tackle that technical debt. And it's really
nice to see the project trying to take that on. It still feels like there's more to be done in
that area, significant work to be done in that area, they're really making it attractive and if you've
thought about plex but you're not comfortable with it for whatever reason take a look at jellyfin
it will have a link in the show notes well let's change gears and talk about proxmox proxmox is an
enterprise class virtualization system based on linux i used to use it in the studio now we run
on arch and just put it all in containers.
But I really like Proxmox and it's developed nicely over the years. And this last week,
they announced the Proxmox Backup Server. This will be an enterprise grade backup server for
your virtual machines, containers, and the physical hosts. It is specifically optimized
for the Proxmox
virtual environment platform and allows you to backup and replicate the data from that platform
securely. And it has easy management and command line tools. It's got a web user interface as well.
And it's licensed under the GNU AGPL version 3. Hey, those are all good things. Other good things
about the Proxmox backup setup,
and especially on the server side,
is that it supports incremental backups,
deduplication, compression, and authenticated encryption.
Plus, Chris, you'll love and uses Rust as the implementation language
so that there's good performance and low resource usage
and maybe a high-quality code base.
Time will tell on that one.
It also has strong encryption on the client side.
And hey, that's where you want it.
So you can back up to something you might not fully trust.
What I love about Rust is it's used to sell a tool as production grade
just because you name drop Rust.
Have you noticed that?
Yeah, I don't know that that's really fair
because any new code base is likely to have bugs,
even if you have an amazing compiler
like you do with Rust. Yeah. But then again, if you were to come to me and say, hey, Chris,
if you want us to build some really nice, high performance, low level enterprise grade tools,
what language do we write it in? I'd probably say Rust. And here I thought you liked C++.
All right. Well, that concludes the Rust Watch for this episode. seen in general, making backups easier for folks, right? If you're going to have to go do it yourself or learn a whole bunch, while that might be valuable, it might also mean delaying taking
those backups. So if you can just integrate with, you know, with these new features, that's great.
Yeah, I wish we had a reason to use it here. Like if that Archbox would screw up a whole bunch,
I'd say we should Proxmox it up. If I wasn't all pied up in the rv and went x86 i'd probably use proxmox underneath whatever
i ran but i just haven't had that use case for it yet and the reason why it kind of bothers me is i
want to talk about it more on the show because it's really good and we just don't talk about it
enough but i just haven't quite found that use yet. Before we kind of changed everything up and went on a free
NAS tear for a while and all that, we used to use Proxmox to do a lot of what we now do in containers.
But it just made sense to use containers for the types of applications we're doing. If you have a
good use case out there, let us know. Linuxonplug.com slash contact. All right, now for the Microsoft Watch.
Those crazy cats at Microsoft have released ProcMon.
No, no, no, not that old SysInternals one.
ProcMon for Linux.
It's a process monitor tool, and it's in preview now.
They write, process monitor is a Linux reimagining
of a classic ProcMon tool from SysInternals suite
of tools that were for Windows.
ProcMon provides a convenient and efficient way for Linux developers to trace the syscall activity on the system.
And I have to say, I was shocked to find it is not in the AUR, at least when I checked yesterday.
Wait, what? How is that possible?
I don't know. I don't know.
Not enough Arch users over at Microsoft, I guess. I mean, they do have a PPA. Yes, they do. They do. So what do we think
about this story? Is it still a big deal when Microsoft is releasing some of these classic
tools as open source on Linux? It feels to me like kind of routine at this point. And I think that
means things have really shifted. You know, this would have been huge, massive news, what, two years ago, but just feels like a regular week with some new stuff from Microsoft. So that's kind of how I feel about it.
Neil, what about you? What about seeing some of these classic Windows tools that were honestly some of the gems of Windows coming from Microsoft to Linux. I started as a Windows developer, and I did do Windows systems administration as well. So like
these tools were things that I relied on when I did a lot more Windows stuff when I was younger.
It's nice for me to see that these are coming to Linux. I personally relied a lot on ProcMon
when I did Windows stuff. So when I saw it was released, I tried to grab
it and build it and package it up for Fedora. But alas, it actually doesn't build. There's
a couple of things wrong with it. They have bundled copies of libraries that don't compile,
and it tries to fetch stuff from the internet, which is a no-go for shipping it in any Linux
distribution. I know that a lot of people think this is kind of the good way to do things now, but like it feels real bad, especially from a company like Microsoft, who
should at this point know the value of reproducibility and stuff like that. And this
also kind of showed a little bit of like they don't really get how to make open source software
that people can easily consume because they
made it so that it's just like a bunch of silly putty thrown over the wall.
Yeah, I think it's pretty early days.
You know, if you look at the commit, it does seem like it's something developed maybe internally,
not introduced into any sort of build system and just now it's out there.
And Neil, is that criticism fair wholly?
Like if you balance expectations with
how large corporations work, maybe this is the best it's going to get for jumping code over the
wall? Because think about it from this standpoint. This is probably after a long series of discussions
about this strategy, this tool, how this developer will spend their time. And there's even like legal
analysis, right?
Like a complete version of the project to an extent that has to probably be reviewed
by several parties before they can release it.
Sure, sure.
No, that's all good.
Well and good.
But Microsoft is not new to this whole trying to release open source software thing.
Microsoft today is not a company that I would consider completely unaware of how you do free software
or open source software, whatever term you'd like. And they continue to struggle with this
in frustrating ways with.NET, with MSBuild and Monodevelop. They're not new to this.
This is something that I expect from somebody who's really new to this and has no experts or no people from the community.
Yeah, I think it's a fair criticism on how the code is released and that they've had time to adapt.
So I follow you there.
Let's talk about something that I think those of us on this show were pretty pleased to see, although we'll wait to see how it ultimately plays out.
But it's official.
plays out, but it's official. It's been approved for Fedora 33 to switch the desktop variants default file system to ButterFS. The Fedora engineering and steering committee formally
signed off on this, which means we're going from extended for to ButterFS. Oh my God. Yes.
Having it formally approved, all the T's are crossed and all the I's were dotted.
We're now, you know,
we're just proceeding forward and getting everything working. Crossing my fingers,
we'll actually have it in beta and GA. I'm pretty confident it'll be at least in beta.
I have no reason to believe that it won't make it to GA. So I am, I'm really happy about this.
And man, years and years of effort are finally paying off.
Well, it is very exciting and congratulations.
And I thought it was sort of driven home recently because we just got an email from Stephen or Stefan.
Sorry, Stefan, if I'm getting it wrong.
He was upgrading from OpenSUSE 42.1 to 15.2.
And he said ButterFS and snapshotting were a total lifesaver. He writes, I've never
had to start over on my main machine. Anytime I hose the system by making a big error or an
upgrade went awry, I just reboot into an older snapshot and use Snapper to roll back once I'm
in a working system. Of course, the implementations would eventually be different in Fedora once it got to that point. But just think about how critical a snapshot and rollback option is
for a distribution that is updated as frequently as Fedora and not just major releases, which are
frequent, but significant package updates, including new versions of the kernel come down
while you're running a current release of Fedora.
And while it generally works really well and DNF is pretty great at catching stuff,
there's nothing like that belt and suspenders where you also have the reliability of those snapshots. And I think it makes the risk of upgrading to the next release of Fedora
even less now. So I'm excited to see where this goes.
Yeah. Right now we're not doing the automatic
snapshotting stuff. That's down the road. That's down the road. I'm trying to be methodical about
how we move forward and integrate these things. Well, it's kind of like I said recently, you have
to get the file system shipping before you can start building the tools based on it. Exactly.
And like we can take advantage of the lessons learned from OpenSUSE, from other companies who
have been rolling it out into production on NAS devices, in server fleets and stuff like that. And we can hopefully
make better decisions and optimize for our use case a lot better with all of this, you know,
knowledge and experience. And that's what I'm hoping to go towards. Like some of the choices
that are a little bit weird from my perspective on OpenSUSE for their layout, we're probably not going to do those on Fedora because we don't need to.
And we're making already some enhancements are coming down the pipeline. There were some patches
written for Libvert to better handle the creation of virtual machines on ButterFS, making them no
cow for the disk image files. So that'll happen automatically. So that means that you won't be impaired by ButterFS
just out of the gate in that sense.
And there's other things that we're working on across the board,
things for like improving performance
when dealing with SQLite, Postgres, MySQL, all the things.
So like we're looking at it,
we're actively doing these things.
I'm confident it's going to be a great experience
by the time Fedora 33 rolls out.
It won't be like super fancy. The idea is that you won't notice anything, but it'll be there
making things better. That's what I'm so excited about is that, you know, suddenly without the
user having to do anything, you've got all of these additional things like copy and write snapshots,
file system compression, optimizations for SSDs, built-in RAID. You might not need any or all of these, but they're all at your fingertips now,
and you don't have to know about it before you go install your Linux system. You can opt into it
later. Yeah, I look forward to trying it in Fedora 33. To tell you the truth, I think it's going to
make it an exciting release just because of this. Even though nothing really fancy is going to
happen, and it's pretty much transparent to the user, I'm still looking forward to it.
And Wes, while we're shucking and jiving about Fedora, one of the things we liked about the previous release was they shipped a user space out of Memory Killer, which in our testing actually made a significant difference if you're on a Linux box with some memory pressure.
And now it seems that this is going to be built into
SystemD. Yes, you guessed it. Because of course it will. Now, I believe this is actually some of the
original code that was developed by Facebook, right? Yes, SystemD, O-M-D is the out of memory
demon developed by Facebook. and now systemd developers.
They're aiming for this to be better Linux handling of out-of-memory or just low-memory situations.
And while Facebook originally wrote this code for their servers,
it's continued to be refined and adapted
so that now it works pretty well on desktops too.
Now, Neil, do we know, does this likely mean
that Fedora will stop using their version and switch to this?
I've been keeping a close eye on this. As I told you, when we were putting early OOM into Fedora will stop using their version and switch to this? I've been keeping a
close eye on this. As I told you, when we were putting early um into Fedora last release cycle,
we were actually also tracking this. This came up as part of it. Things are actually moving faster
than I expected. Like I originally expected that systemd umd would not land for another eight
months, and it looks like it might land in a
few weeks. So this means that we're probably going to start looking at it in Fedora 34.
That makes sense. Wes, one of the things that's kind of neat about having it down there at the
systemd level is it's also cgroup-aware. Yeah, systemd omd pulls systemd for out-of-memory
enabled cgroups to monitor them and then kill based on memory pressure, which is that new measure that Facebook helped add to the kernel or swap usage.
So I think it'll be interesting to see, you know, if this solution that is already in place or this new systemd-based solution, I wonder if they might find different use cases that work a little bit better.
Especially, you know, it does say this works well for the desktop.
find different use cases that work a little bit better, especially, you know, it does say this works well for the desktop, but in the Facebook world, you know, they could opt into all this
stuff, make sure that all their utilities on the server side were playing nicely with C groups.
It may vary a little bit on desktop environments, how well all this stuff works.
Why don't we stop now and do a little housekeeping just to change it up a bit,
because that was a lot of information to go through.
I'm out of memory. Yeah, and we're going to change it up a bit because that was a lot of information to go through. I'm out of memory.
Yeah, and we're going to kill it right now.
And I want to get the word out about Guadalcanal 2020.
It starts July 22nd, which is tomorrow as we record this,
and goes to the 28th of 2020.
Guadalcanal is the GNOME community's largest conference,
and it's online this year.
So I would imagine this could be the first Guadalcanal I ever attended if I had given it any thought.
And maybe I still can.
I'll look into that.
And you can, too, with the link in the show notes.
Also, if you want to join us live, you'll find out when.
Did you know that?
Calendar.
jupiterbroadcasting.com slash calendar.
Link for that's in the notes as well.
And we do it over at jblive.tv.
Every Sunday, the Luplug, the folks in the Mumble Room, get together and hang out
at noon Pacific. That's the same bat time we do this show. It's just on a bat Sunday. We hung out
last Sunday. I talked about how I reflashed the computer in my RV for more performance, and it
made for some good stories. Lots of good stuff. We had some new folks show up. Of course, as always, we talk a little butter FS.
So the love plug is it's often a multi-hour event, but you can show up for as little or as much as
you like every Sunday at noon Pacific. It's also on the calendar if you just want to check that.
So we'd love to see you there. And I've made it easier than ever to get the mumble server info.
I've made it easier than ever to get the Mumble server info.
You just go to linuxunplugged.com slash mumble,
and it's all there, the quick stuff.
Or there's a link at the top of the page.
And the extended guide is still available at the Jupyter Colony page if you want the full thing.
If you're brand new to Mumble and you want a guide
on how to set up all the audio interface stuff,
the full guide's still available.
But if you just need the quick server info,
linuxunplugged.com slash mumble. There we go. See that? Doesn't that feel better, Wes?
So much cleaner in here. And did you put an air freshener in the studio?
Yeah, that's just Levi. I gave him a shower. Yeah, he's much better now. So let's talk about
Element. This is something we've been watching really closely.
The Riot Chat program and the company actually has been renamed to Element, which is a matrix chat client.
Element, which I don't know if that's any easier to Google than Vector.
I guess we'll see.
It's the new flagship secure collaboration app for the
decentralized Matrix communication network. Element lets you own your end-to-end encrypted
chat server while still connecting to everyone else in the wider Matrix network. Yeah, that's
right. That's the big federation play. We did know this name change was in the works.
They explained it even a few weeks ago, mostly in a blog post, talking about some of
the issues they had with gigantic game companies who blocked them from being able to trademark the
name, which then had a knock-on effect of them being unable to go after abusive forks in the
Google Play Store. And it just essentially came down to also being associated with violence rather
than a more constructive form of chaos
that they're into
because they write on their blog.
And I think that's nice
because the name doesn't really matter in the end.
I got used to Riot.
I'll get used to Element.
And I think it's great
because they've also renamed the company now
to just sort of keep it simple.
Right. It's a little more consistent.
And you're right.
We just got to get used to these things.
You got to put a label.
And naming is hard.
Don't we all know it?
It's the worst part
about coming up with a new podcast is coming up with the name. It's the
worst. They said they wanted to future proof it as well. So this was their rationale for it. They
said as matrix.org announced last month, P2P matrix is in heavy development and we see a world
where element will literally be an element of matrix running your own home server within the app. So you can communicate
if you don't have a server or even internet. That actually is pretty cool. That is really cool. And
that's one of the things, you know, the whole matrix ecosystem is filled with fascinating ideas
and a lot of potential. The math names, I kind of love too. And they're not just doing name changes,
Wes. They're rolling out pretty big improvements on their mobile apps and a massive refresh to their web UI, which looks really good. I am seriously eyeballing Matrix right now as the chat platform for Jupyter Broadcasting.
JupiterBroadcasting.com slash Telegram. We haven't plugged it a lot recently because I've kind of just been watching how the chat ecosystem is unfolding. Because for Self Hosted a proper fit with this show, I think something like Matrix, which interconnects to all the different platforms, which is the kind of thing our audience can geek out on.
But also it's free and open source and it's self-hostable.
All these things sort of stack my general bias towards Matrix as a chat platform that would potentially replace IRC and Telegram.
And we would just use this for all of our primary text-based communication with the community.
But at the same time, there's that network effect to consider.
So I have to look at things very seriously like IRC, Telegram, and Discord.
And so I've been keeping an eye on Matrix for a while
and sort of mentally doing the math on how it would play into a wider ecosystem for us.
And with Element now making these improvements
to the web UI and the mobile clients,
you have the general connected nature of Matrix
and this new P2P stuff they're working on.
I find that to be a really compelling package all in.
We need it to be easy, frankly.
You know, it's got to be worth the install. So the client can't be gross. It's got to be easy, frankly. It's got to be worth the install. So the
client can't be gross. It's got to be easy to connect with us. When you live in a world of so
many solutions, that's how you differentiate. So Element's a thing now. No longer will it be called
Riot. Now, I don't really have much to say about this last one. It's just a little weird.
little weird. There's a job posting over at Microsoft's jobs page about a program manager for a cloud PC project that would be a desktop delivered from Azure and managed by Microsoft
at a flat per user price. They call it the Microsoft Cloud PC, a new offering that would be built
on top of the Windows Virtual Desktop to deliver desktop as a service. It would really be targeted
to business customers that are looking for a modern, quick workstation experience that have
high connectivity. And I have to tell you, I could see this being appealing because I know of companies that sell versions of this based on AWS right now.
And you go into the businesses and they don't have any on-premises services other than a switch, a router, and a Wi-Fi access point.
There's no file server.
There's no directory server.
It's just a flat network that provides high-speed internet access. And I think you can hear that in the first sentence of that description that you read there,
that this is the Azure and Microsoft managed solution, that this already exists.
And yeah, clearly it's interesting because as anyone who's administered Windows PCs knows,
it's kind of a pain.
And if you can just let Microsoft handle it,
if you already have to be an internet-connected company anyway
from all the other SaaS products that you buy that makes your company run in 2020, yeah, all right, why not use the desktop?
And maybe, maybe Microsoft can do this better than others since they control all the pieces.
What's funny is I know that Linux has all of the pieces to do this because I rocked Linux terminal servers way back in the day.
And we have all kinds of
means to accomplish this now. And there are companies out there working on it. But where
Microsoft is going to have this sort of home turf advantage is they're going to sort of package it
all up with Microsoft's Office 365 offering, which it's sort of like this obvious solution. You have all your data in the ecosystem
of Office 365, Word, Excel, Teams, OneDrive is your storage or Teams, it's weird. And you just
need essentially a Chrome PC. But instead of getting that Google stuff that's all connected
to the Google ecosystem, you get Microsoft's version of it. And this one's a step further in its maybe even a more pure version of that Chrome experience,
where it's a complete remote cloud PC managed by Microsoft as part of your Office 365 subscription.
You know, what I want to know, is WSL going to work on cloud PC?
Brent, I'm sure you'd get one of these, wouldn't you? Well, it means maybe I need less
fancy hardware all the time. But no, I'm sorry. I'm not in. My hand is not raised. I would like
to see a vendor come along and offer this up with Mate as the desktop experience, LibreOffice as the
office solution, connect it to some sort of central cloud storage so that way you can have multiple users from a workspace use the same storage, maybe create some sort of user ID system for them.
I mean, there's small versions of this have been implemented or there's Amazon's implementation of these, but there's not a complete package of you need Office and a computer and you just need these storage services for your company.
And here's another alternative for you that's not Google or Microsoft.
The business ready, all wrapped up, ready to go stuff.
Yeah.
And all of the pieces are there.
And it's something that Linux could be particularly good at because I don't know which system you'd want to use, a remote desktop system or a Linux terminal services setup.
But the cool thing about a Linux terminal services setup is a lot of those programs run locally in memory,
and they will use your local GPU if you have one, which means you can make thin clients that
have GPU acceleration. But additionally, if LibreOffice is loaded in your local RAM,
your internet can blip or your latency can be bad and you're completely unaffected on a Linux
thin client setup. Now, if you're doing something like XRDP or another type of remote desktop
protocol, that wouldn't be the case. So it just depends on the implementation, but there's so
much opportunity here. It makes me wish I had the time to like start a business because there is a
whole area of small business in the States that is somewhere between five to twenty five people.
And these annual subscriptions just stack up and they're a little allergic to them and they just
need somebody to show them a solution that works for them and give them an implementation.
That's not some sort of moral choice. They're not choosing Microsoft because of some moral
reason. It's just they don't have any other options presented to them. Right. They have the most handholding that way,
and that's what they need. But kudos to Microsoft. I think it's probably going to be a successful
product if they launch it. And they've tried stuff like this in the past, but I think,
you know, now is the time. I wonder if it's, you know, we've kind of talked about other Microsoft
assets like Office 365, you know, things you used to have to have a Windows desktop to use,
or maybe a complicated wine setup that since they've migrated to the cloud into an online
offering, while we might not love having to use them from the free software world,
it's good for people trying to use a Linux desktop and still get work done.
And I mean, this almost seems like it's the ultimate in that it's making Windows the same
thing. I somewhat recently had to go, for an unfortunate reason, go spin up a Windows VM,
and I did it on Azure because I didn't want to clutter up any of my local systems. And it was
kind of a painful process with all of the pains of using Windows. If I could just have a cloud PC,
pop it up, get my Windows-specific stuff done, and then go back to my pure Linux world,
yeah, all right, I'll pay Microsoft a couple bucks for that.
So do you already pee into an Azure Windows VM?
How does that work?
They've got like a web,
a way to view it on the web too, of course.
But yeah, I just RDPed up in there,
got my work done and got the heck out of there.
But it took forever.
And of course, you know, I was doing updates
and had to spin all the way up
and getting it actually installed and booted up
took much longer than an equivalent Linux VM. I was honestly surprised by how painful it was.
Huh. That's an interesting experience, Wes. Thanks for sharing that with the class.
All right. So if anybody out there knows of a service like this for Linux, where you could go
set up an account, it'll spin you up a virtual desktop, and then you can remote desktop into it,
let me know, linuxunplugged.com slash contact
or chrislass on Twitter,
because I'll give that a try
and I want to give it a review
because that's a product that should exist
for Linux desktop as well.
We can do it better, maybe.
RDP is actually kind of legit,
but we can use XRDP.
So we could probably do it better.
What do you say we do some picks
before we get out of here?
Because we got some good ones
and we got some feedback too.
And we start with Polybar, a fast and easy to use status bar and i think a couple of
you've had some experience with this one yeah well actually i just stumbled across it the other day
and i was thinking to myself man this would be a great addition to something like openbox
or i3 and then after mentioning it to you guys i believe drew didn't
you try it on i3 yeah i've used it a number of times on i3 i'm not using it right now even though
i'm on i3 but it is a really great and good looking bar that's easy to get going and easy to
add like custom blocks and stuff too and change the way it looks and give it a nice color scheme.
And all around, it's just a little slicker and more fully featured than, you know, the stuff that like comes stock with i3.
The orange one's been using it for about a year.
So give us the why I've been using it for a year review.
Originally, when I started with i3, naturally, I found a rust implementation of those bars,
and it's really nice.
But the nice thing about Polybar
is there's just tons and tons of extensions.
And the fact that you can have it show Spotify,
CPU usage, volume, time, current window,
everything is just great
and massively beats the Windows taskbar and what you'd
normally expect from a taskbar like that hmm that's a pretty hearty plug so check out polybar
we'll have a link in there and then we have a fun one for those of you who like your stats it's
called y top and it's essentially go, only this time it's in Rust.
And you thought we were done with the Rust watch. Heck no.
Yes, another system monitor, only this one is written in Rust. It's familiar to those of you
who have been enjoying Bashtop. I think we mentioned that in episode 351. So it's been a little bit.
But this is another one.
Now, there's Go Top out there as well.
But this is Y Top.
And yes, it's Rust-based.
You know, I think our unofficial goal here on Linux Unplugged is to just find every single possible top implementation out there and talk about them on the show.
And I'm interested in what each one chooses to show.
Because, yeah, okay, you're all just going to go scrape the kernel for a whole bunch of information and display it for the show. And I'm interested in what each one chooses to show. Because yeah, okay, you're all
just going to go scrape the kernel for a whole bunch of information and display it for the user.
But how you display it, that kind of counts. And you know, Ytop, it's a decent contender. It's got
a really nice CPU graph going. I also like the network usage graph that it displays because,
hey, right now I'm chatting with you. I can monitor my connection.
Yeah. And the hit on the system seems to be very minimal,
which is always my thing.
I don't want the monitoring tool to put abusive load on the box,
which is really actually kind of an issue on the Raspberry Pi. You have to be careful.
Like on the Raspberry Pi 3, you could run net data,
but it basically will start consuming the device,
especially if you're running it off the SD card.
But something like this is so much lighter.
And if you just need quick information over a terminal session,
it's hard to, a bash top might look a little better,
but it's, I don't know, it's hard to beat this one.
I like it a lot.
So get a link to those in the notes
and feel free to send yours along.
Why don't we get to a little feedback?
Wes, do you want to take this first one about some NextCloud follow-up and questions? Oh send yours along. Why don't we get to a little feedback? Wes, do you want to take this first one
about some NextCloud follow-up and questions?
Oh boy, Eric, a pedestrian,
had some feedback for us
about our recent discussion around NextCloud.
Hello, Chris and Wes.
I really enjoyed the NextCloud discussion,
but I have a few questions and comments to share.
One, are you using Redis?
If not, it might go a long way
towards addressing any concurrency issues
you're still seeing.
Yes, we are using Redis,
although I don't know that we've tuned it a whole bunch.
We're using a Docker Compose stack
set up from the Nextcloud project itself.
So we haven't spent a ton of time
going outside those bounds,
but we are running a Redis container.
That's a good tip though.
Number two, using the
S3 compatible storage background has been great for me. The interesting part is that any S3
compatible storage provider will work. I've been using a DO droplet with Wasabi as the backend,
and it's both cheaper and more performant than Spaces. That's also probably a good tip. You know,
we went with Spaces because, well, we were already a DL customer. We had access to it.
We had accounts set up for, you know, access to the team, all that stuff.
So it made sense to plug right in.
But the whole beauty of using something that is S3 compatible
is you have options there, including S3 itself, of course.
Now, I've not used Wasabi,
but I will say we've received a couple other items of feedback
that are also recommending it.
Yeah, like Rick here writes in, and he said in your recent episode when you talked about
your expensive NextCloud implementation, I found it totally unacceptable.
I cringed when I heard the price.
This is homework for you and the JB crew to do.
He says 10 terabytes of hot AKA active storage on Wasabi's service is $59 a month. Combine that with the 15 a month droplet,
and it should be just about 80 a month with tax. Overages, if any, would be much more affordable
and manageable. Can NextCloud integrate with Wasabi? Hell yeah. Here's what I found to get
you and the crew started. And he says it also allows you can pick a CDN in front of Wasabi. He says, I have an idea.
Use the storage on a droplet for temp storage and then move it to Wasabi instances as needed.
You just figure out a storage strategy and create docs for the crew. Save your hard earned money. I
agree with that. We are going to fix it. For the record, I'm not affiliated with Wasabi or any
storage company. I think this is something we'll have to look at
because we're either going to have to stop using it
or switch to something cheaper.
And if this math is right, I mean, around $80 a month,
it's still more expensive than Dropbox,
but not by much,
especially if you consider how many users we have
because Dropbox charges by the user.
Right.
If you consider how many users we need, our cost is probably around $100 a have because Dropbox charges by the user. Right. If you consider how many users we
need, our cost is probably around $100 a month for Dropbox if you were to take the annual cost
and spread it out. So $85 would be cheaper. It's still a lot of money. It's still a lot of money.
And I wonder if what we really need is something that's just a small, tiny storage of in-progress
files. I think that's a big part of this is we got to go,
we should go do some pruning and figure out like,
what do we need to have hot and what can be left in archival storage?
And then invest in storage at the studio in the physical box here that we can add to.
But then it just sort of changes the backup scenario here in the studio.
So I got to think about that.
It's a can of worms, but we do appreciate the feedback, guys.
It's given us something to think about.
Now, Wes, there was somebody who wrote in asking for timestamps.
You want to take that one?
Oh, yes.
Okay.
I guess this is a request for me then.
Advait writes in, wanting timestamps.
Here's a humble request that timestamps be included in the podcast description
so I can go right to the segments I most want to hear.
Thank you.
Love the show.
Great work.
Okay.
Hey, that's pretty reasonable.
We do have them, you know, available as chapter markers,
and we've got them on YouTube, but that might not be everywhere.
I guess that's something we could consider
because I guess we could just extract it from the chapter information
and just put it in the description. Yeah, we can add it to the long form description. That shouldn't be too
big a problem. That's something we'll think about. But this also is just a good chance to remind you
that we do have chapter markers for almost all our shows, all our shows. Yes. And you can jump
right past any segment you like, or you can jump to a topic using those. And they should be in just about every mobile
podcast player. And we'll start looking into embedding it in the show notes as a possibility
to there because we already have the information. All right. Now we have one last feedback that I
will read. And it's a plug for a audience member's own project, but it's so dang cool.
I wanted to talk about it.
He said, I've created an app just for Linux that I think fellow space nerds like you would be interested in.
It's called Astro Ninja, and it's written in Python using mostly PyCube 5 and Scrappy Spiders.
I love it.
It has the full worldwide launch schedules, industry news, Hubble images, launch tallies, and SpaceX live streams built in.
The project just reached its alpha release, and I'm trying to invite like-minded individuals to check it out.
And seeing how I really, oh, geez, now he's bragging.
Now he's talking nice about it.
He loves the show, and he says he's been listening for years.
And he said he first thought of us and wanted to pass it along. So in the show notes, we'll have a link to his email where he has links to a YouTube video that shows it off.
And we'll put a link to his GitHub repo, which you can go check out the project if you're a space nerd, want to geek out on that stuff.
It's Astro Ninja and it just hit alpha and it's up on GitHub.
Kind of cool.
This looks pretty handy as a little dashboard and and it's not another Electron app,
so you just need Python and Qt installed, and away you go.
Yeah, most Linux desktops are going to have that.
So thanks for sending that along, Tom,
and congratulations on the release.
It does look like it's a pretty cool project.
AstroNinja is a good name too, right?
I love that.
I love watching those SpaceX launches too.
Right? And you know, they use a bunch of Linux. So clearly, Linux and astronomy,
they're just made to be together.
All right. Well, that brings us to the end of this week's official program. Thanks so much
for listening. We'd love to have you tune in live over at jblive.tv. Or if you can't,
thanks for listening on the download. We like that too. That's just fine. In fact,
why not share this podcast with a Linux-loving friend?
That'd be great too.
You can find links to everything we talked about at linuxunplugged.com slash three...
63?
63.
Yeah.
Thanks, Wes.
I was going to get there.
363, linuxunplugged.com slash 363.
And the show is at Linux Unplugged on Twitter.
Thanks so much for tuning into this week's episode of the Unplugged program.
And we'll see you right back here next Tuesday. So, ByteBit, you were chatting in the IRC that there is a Linux desktop service out there called Ubi Desktop.
Yeah, so I stumbled upon it, and it is kind of neat. single core 3.4 gigahertz, 2 gigabytes RAM, 32 gigs
of SSD storage and 1
terabytes a month of traffic
at 1 gigabit a second.
How do you know how much your desktop
traffic's going to... That's a tricky one, but a terabyte
you'd think would be enough.
Yeah, and
it is dependent on if the
RDP or what kind of
VNC kind of service, if that data gets counted against it or not.
But it is very easy and they have a free to try option.
And if you scroll a bit more down, they have a discount coupon code.
And they have an extreme version.
And they have an extreme version at $43.90 a month.
You get four virtual CPUs, eight gigs of RAM, 128 gigabytes of SSD and five terabytes of traffic.
That SSD is a little tight, though.
That's interesting.
If you were a new business that was starting up, you didn't have any machines, right?
And someone was going to sell you this service.
Would they just run on thin clients?
Yeah, exactly.
I mean, think about another way. Isn't it silly to buy $800 PCs with $125 monitor and a mouse and keyboard that you then go license software for, then you have to have an IT person
manage it? How silly is that for an office of five people or 15 people or even 100 people?
And if you scroll down far enough, do you need a VDI?
Try desktop as a service.
They even support those kind of things.
Yeah, it's pretty wild.
So it's coming.
And obviously, it doesn't mean it's going to replace traditional computers because there's
so many reasons why you still need a traditional computer at a desk.
But you got to figure if Chromebooks have been pretty successful, there's probably a market for this kind of thing too. And I've heard from listeners in the past
that use this as a daily driver as part of their work. It's some other vendors virtual desktop
solution right now, but it's already out there to some degree and being widely used.
One challenge I can't get around is the necessity for always connected internet.
I run into this all the
time, even at home, you know, okay, I've got this stable connection and it just, you know,
drops. And so if I'm in the middle of something that is one of the biggest productivity killers
ever. That's why I think it's so office focused. You know, you picture someone in a downtown
office suite, their building has fiber connection, probably, you know, the best case scenario.
It's basically the opposite of wherever Brent is.
Yeah. Or me.
Yeah.