Werner changed the topic of #armbian to: armbian - Linux for ARM development boards | www.armbian.com | Github: github.com/armbian | Commits: #armbian-commits | Forums Feed: #armbian-rss | Development Talk: #armbian-devel | Type 'help' for help | Logs: -> irc.armbian.com
<archetech_n2> libreoffice that pig has to go otherwise looks good
<rneese_> ?
<rneese_> well the goal was a full working desktop
<rneese_> thats why he adds it
<archetech_n2> who's he?
<rneese_> that one is IgorPec
<archetech_n2> figures
<rneese_> well I spent today resting tomorrow i will get back to the desktop builder
<archetech_n2> update-passwd has found a difference between your system accounts and the current Debian defaults. It is │
<archetech_n2> │ correctly. For more documentation on the Debian account policies, please see │
<archetech_n2> │ /usr/share/doc/base-passwd/README.
<archetech_n2> │ advisable to allow update-passwd to change your system; without those changes some packages might not work │
<rneese_> but once we have it working I would like to see a website where you select
<archetech_n2> is this normal? never seen it before
<rneese_> never seen it
<archetech_n2> he proposed change is: │
<archetech_n2> │ Change the home directory of user "irc" from /var/run/ircd to /run/ircd
<archetech_n2> │ │
<archetech_n2> seems its misconfiged
<archetech_n2> it dont like this user irc:x:39:39:ircd:/var/run/ircd:/usr/sbin/nologin
<archetech_n2> I will allow thechange
tehircz has quit [Ping timeout: 246 seconds]
<archetech_n2> zip: stdout: No space left on device
<archetech_n2> E: mkinitramfs failure cpio 141 gzip 1
<archetech_n2> hmm first update attempt
<archetech_n2> df -h shows 12 of 15gb free
tehircz has joined #armbian
<archetech_n2> Errors were encountered while processing:
<archetech_n2> E: Sub-process /usr/bin/dpkg returned an error code (1)
<archetech_n2> initramfs-tools
<archetech_n2> ill try reboot
archetech_n2 has quit [Quit: Leaving]
<rneese_> hmmm
<rneese_> you on emmc or sd
<rneese_> i hurt going to bed talk tomorrow
sunshavi has quit [Remote host closed the connection]
<archetech> cya you can address this when fresh
rneese_ has quit [Ping timeout: 260 seconds]
<archetech> because /boot partiton is full and it cant build things
<archetech> breaks all updates
<archetech> lanefu: did you drop this bomb on me knowing its broke ?
<archetech> ie had you installed it or one close in date
<nekomancer[m]> ho-ho. who set 80 mb default /boot size?
* nekomancer[m] set it 200 mb when make non-ext image
<nekomancer[m]> cost-cutters
<nekomancer[m]> archetech: just move storage to another system and resize /boot
Ntemis has quit [Read error: Connection reset by peer]
sunshavi has joined #armbian
ChriChri_ has joined #armbian
<archetech> storage?
<archetech> its the whole OS partition
<archetech> but yes thats the only way to resize the 1st partition
<archetech> this is kinda amaturish oversight imo
ChriChri has quit [Ping timeout: 272 seconds]
ChriChri_ is now known as ChriChri
<nekomancer[m]> o, it's just auto-expand not occurs?
<nekomancer[m]> then just run `cfdisk` or `sfdisk` (i forget who) from root on that system
<nekomancer[m]> then in terminal GUI choose partition then resize
<archetech> I see this uses btrfs is this your base creation with rneese xfce on top?
Tonymac32 has joined #armbian
<nekomancer[m]> btrfs means sepereted /boot partition
Tony_mac32 has quit [Ping timeout: 256 seconds]
<nekomancer[m]> separated
<nekomancer[m]> and yes, by deafault there only ~12 mb free
<archetech> I see this uses btrfs is this your base creation with rneese xfce on top?
<nekomancer[m]> there can't be 2 copy of initrd
* nekomancer[m] not know.
<nekomancer[m]> rneese seems gone to bed
Tonymac32 has quit [Read error: Connection reset by peer]
tehircz has quit [Ping timeout: 246 seconds]
tehircz has joined #armbian
drobo_01 has joined #armbian
drobo_00 has quit [Ping timeout: 265 seconds]
drobo_01 is now known as drobo_00
DaRock has joined #armbian
sassinak-work has quit [Ping timeout: 244 seconds]
sassinak-work has joined #armbian
xecuter has quit [Ping timeout: 272 seconds]
xecuter has joined #armbian
tehircz has quit [Ping timeout: 260 seconds]
tehircz has joined #armbian
tehircz has quit [Ping timeout: 260 seconds]
tehircz has joined #armbian
Tonymac32 has joined #armbian
<lanefu> archetech: yeah btrfs root, xfce desktop
<lanefu> archetech: i ran into that a bit, there's some post-hooks that are conflicting
<lanefu> not sure why
<lanefu> basically debian's tryign to do some dumb x86-style rootfdisk images, which arent actualy being used
<nekomancer[m]> just tiny /boot by default. allows only one unsafe update scenario: remove old content, then create new. If anything goes wrong — you get unbootable image.
<lanefu> yeah tk was pretty opinionated about /boot size.. it used to be 50 megs
<lanefu> but.. this is actually a packaging issue
<lanefu> basically when kernel updates work correctly.. everything gets uninstalled cleanly, then new packages are writen
<lanefu> all is well
<lanefu> but if you do like dpkg -i to install a new kernel can make a mess
<lanefu> and in this case, i think the kernel i had built was missing normal post-hooks that make debian do different thigns iwth the image contents
<lanefu> archetech: if you "freeze" the kernel in armbian-config
<lanefu> i think your upgrades willbe clean
sunshavi has quit [Remote host closed the connection]
<archetech> poss but im building my own images now
<archetech> for QC reasons
<nekomancer[m]> archetech: then just set boot size in command line. easy and reliable
<archetech> whoever thinks 100 MB is good size is plain wrg when min of 15gb is avail on avg user sdcard
* nekomancer[m] uploaded an image: 8 Hoozuki no Reitetsu - 08_26.webp (46KiB) < https://matrix.org/_matrix/media/r0/download/matrix.org/mDFicFeNAIPiNLdhWpxWRqZb/8 Hoozuki no Reitetsu - 08_26.webp >
<nekomancer[m]> ooops. sorry.
<lanefu> :P
<nekomancer[m]> 100 mb still not enough. at least 2× 86 mb should be to allow double boot set
<nekomancer[m]> I propose 200 MB
<lanefu> for historical reference https://quinled.info/2018/09/15/quinled-dig-uno/
<lanefu> wrong link
sunshavi has joined #armbian
<lanefu> anyway that provides background around the last time we expanded
<lanefu> archetech: anyway sorry the image wasnt as graceful as we'd like
* nekomancer[m] think «remove first» strategy is generaly wrong.
<lanefu> yeah.. anyway 100 had been okay most of the time
<lanefu> this is being caused by extra initramfs images in a format that we don't use with uboot being stuffed in /boot
<lanefu> and i don't know if thats just its bullseye, or what
<lanefu> will have to test with another distro
<archetech> whats a easy section to disable in kern to make it smaller
<nekomancer[m]> there are tons of drivers
<lanefu> yeah but you're missing on what hte problem actually is
<lanefu> /boot/uImage shouldn't be there
<Tonymac32> correct, it should not
<lanefu> here's me listing /boot on the odroidn2, then 2 other armbian devices http://ix.io/2Bdd
<Tonymac32> it's that shit u-boot
<archetech> yeah where do I change that to lates
<archetech> latest
<Tonymac32> I think the potato and K2 and C2 had to have uimagw
<Tonymac32> well the mainline works but we haven't pushed it out, I gave it a shot some months ago and had a kernel panic for some reason or another I didn't have time to debug
<lanefu> ./compile.sh EXPERT=yes ARCHETECH=latest
<Tonymac32> since the handoff from uboot to kernel seems to be a damned hope and prayer
<lanefu> Tonymac32: are all meson64 using same u-boot?
<Tonymac32> no
<archetech> lanefu: lol
<Tonymac32> everything is mainline except for the N2 and C4
<lanefu> ahh
<lanefu> where we're using ordroid's uboot
<Tonymac32> I mainlined the K2
<Tonymac32> U-boot 2015.whereAreMyDentures
<Tonymac32> OK, I'm over CircuitPython already, it either "just works" by constantly running your "almost ready" code, or it seems to do nothing at all
<lanefu> lol really? that bad
<Tonymac32> I don't know, I spent 10 minutes with it and can't get it to behave as advertised. Just venting
<Tonymac32> lol
nkuttler has joined #armbian
<Tonymac32> disclaimer I'm also running it on an ESP32-S2 board I built
<lanefu> import fix from fixer
<lanefu> fix.makeitwork()
<lanefu> did you do that?
<Tonymac32> rofl
<lanefu> might need that library from archetech's repo
<lanefu> archetech: who you yellin at
<archetech> what is this stuff enabled ive never heard of out it goes
<lanefu> journalling flash file system
<lanefu> i don't know why i have that memoriezed but i do
<archetech> this is ext4 and thats it
<lanefu> used it on wrt stuffs
<archetech> cramfs?
<archetech> erofs
<Tonymac32> that's the kinky file system. better be careful
<lanefu> well luckily the beauty of using the armbian build system is you can easily make custom images as lean as you like
* nekomancer[m] > <@freenode_lanefu:matrix.org> /boot/uImage shouldn't be there
* nekomancer[m] just don't know it
<archetech> ceph? cmon
<archetech> dont nobody care how long these things compile?
<Tonymac32> ...10 minutes?
<ArmbianHelper> .10 minutes? [en~>eng]
* Tonymac32 shrugs
<nekomancer[m]> why N2 have human-editable boot.ini, but rockPi4 — half-binary boot.scr?
<Tonymac32> because the N2 U-boot should have been shot behind the shed 4 years ago
<archetech> plan 9 fr outer space FS
<Tonymac32> boot.scr is standard and supported
<archetech> ok 2 cores of my I5 3570 is on the job
<archetech> heh
<Tonymac32> the readme gives all the details about using the mainline u-boot
<Tonymac32> except for why the kernel implodes, but I think I know why now, we had uimage
<Tonymac32> which is... UGH
<Tonymac32> lol
<lanefu> nekomancer[m]: yeah has Tony mentioned.. odroidn2 is using hacked up BSP uboot 2015 fork by odroid... so its stupid
<archetech> well there fancy boot.ini footwork for that I though
<lanefu> i need to goto school on the image formats
<archetech> for uimage
<nekomancer[m]> <Tonymac32 "because the N2 U-boot should hav"> sounds like to have human-editable boot config file is something bad :(
<Tonymac32> plaintext boot script could be tampered with
<Tonymac32> or accidentally mangled by noobs
<nekomancer[m]> and it is good, no?
<Tonymac32> to weed out the noobs? maybe XD
<lanefu> boot.ini is dumbing down embededded system philsophy for integrity and robustness in the spirit of making it easier for noobs
<nekomancer[m]> or just expand this idea to all linux at all
<lanefu> which is FINE with a vendor image.. but not so fine for us
<Tonymac32> Lanefu, merge window closes today doesn't it
<nekomancer[m]> and start to provide signed encrypted monolytic binary.
<lanefu> Tonymac32: just commit a blank function. then anythin gafter is just a bugfix
<Tonymac32> ROFL
<lanefu> :P
<Tonymac32> commit something completely broken
* lanefu works with enterprise developers
<Tonymac32> fix it later
<lanefu> it's called a feature flag
<lanefu> adding the code and using it are 2 different steps
<lanefu> lol
<lanefu> anyway i'm gonna goto sleep. yo all have a good evening or morning, or whatever time of day you prefer to celebrate
<Tonymac32> lol
<nekomancer[m]> <Tonymac32 "to weed out the noobs? maybe X"> everybody is a noob in some area for some time
<Tonymac32> of course, but I was following with your suggestion
<Tonymac32> ;)
<Tonymac32> the boot.ini in that uboot wouldn't even work from EXT4 until some amazing hacking was done, then amlogic I think fixed that (ext4 was broken in that old of u-boot)
<nekomancer[m]> <Tonymac32 "commit something completely brok"> that will not pass via Igor
<Tonymac32> shhh, don't wake him up and he'll never know
<Tonymac32> it will build
<Tonymac32> rofl
<Tonymac32> I'm a little upset I missed the uimage garbage, I'd have had this fixed in March or so
<Tonymac32> but, spending a couple hours and having to run off to other things will do that
<Tonymac32> this release cycle I wound up fixing borked RK3328
<Tonymac32> which is, disappointing
<Tonymac32> of course maybe that wasn't it and I'll be stumped again
<Tonymac32> we'll see
<Tonymac32> I have a couple hours now
<archetech> one ver of 3328
<nekomancer[m]> rock64
<archetech> not all 3328 boards u fixed
<archetech> ya v3
<Tonymac32> I have no real compulsion to fix a board the vendor put faulty ram on
<nekomancer[m]> oh, that revisions ^(
<Tonymac32> V1 works too
<Tonymac32> and renegade
<nekomancer[m]> my rock64 is v2
<archetech> not faulty ram its lower spec ram
<Tonymac32> some of them work
<Tonymac32> if it doesn't meet the BSP minimum for the device it doesn't matter if it's lower spec, refurbished, counterfeit, etc, that's defective
<archetech> v2 needs a slightly diff ram speed is all fix is avail nobody cares to make a v2 uboot though I get that
<Tonymac32> now the u-boot should be starting at 333 MHz
tehircz has quit [Ping timeout: 272 seconds]
<Tonymac32> unless the mainline ATF/uboot is setting it faster now, but the blobs are sloowwwwwww
<archetech> ok then theres the sdcard coruption on boot too
<nekomancer[m]> Why I can't just tune 2 or 4 numbers in text config on sdcard and forget about this problem until I need new sdcard?
<archetech> yes atf is set fast
<Tonymac32> could be drive levels, although u-boot shouldn't be writing to the SD card, only reading?
<nekomancer[m]> "because of idea of embedded robustenesss"?
<Tonymac32> nekomancer[m] for which? boot.ini?
<nekomancer[m]> kind of.
<Tonymac32> or for the V2 Rock64
<nekomancer[m]> yes, for rock64 v2
<Tonymac32> TBH the rk3328 uboot situation was a circus for a long time, it still is a mess but is slowly getting cleaned up with mainline support
<Tonymac32> every vendor had their own thing, which we simply can't do because of the ridiculousness of supporting 5 u-boots for 5 boards
<Tonymac32> My early on feedback (recurring) is that we should not support anything that does not have either a vendor committed engineer to support their crap u-boot, or mainline
<archetech> nobody but neko bought one of the cheapo ones based on irc traffic
<nekomancer[m]> why that "creators" (i talking about board creators) do own versions of all?
tehircz has joined #armbian
<Tonymac32> cut/paste of BSP u-boot with patches for where they varied from the cut/paste design
<Tonymac32> those patches are usually really hack, so they don't play nice with other boards to integrate them into 1
<nekomancer[m]> <archetech "nobody but neko bought one of th"> I buy this not so cheap earlier than start to read this irc
<archetech> Tonymac32: so you got n2 commit for latest uboot done?
<archetech> was that the link above?
<nekomancer[m]> it was 75€ + emmc + 25€ delivery.
<Tonymac32> lol no, it's been 20 minutes
<Tonymac32> that was the mainline u-boot commit with the how-to
<archetech> but you looking at it now?
<Tonymac32> yes
<archetech> ok
<archetech> just curious
<archetech> nekomancer[m]: ah sry to hear
xecuter has quit [Remote host closed the connection]
<archetech> was thebug then
<Tonymac32> buy some RAM and swap it on the board. :P
<archetech> how u get a kernel in 10 mins 2 cores this will be an hr I bet
<nekomancer[m]> ahhh... so sad... It's not easy to choose good SBC.
* Tonymac32 am trolling, has the setup to do it but has never attempeted it
xecuter has joined #armbian
<archetech> yeah micro soldering no ty
<Tonymac32> archetech I have an i7-3770 dedicated to it with a M.2 SSD, and I force using a RAMdisk
<Tonymac32> overclocked to 4.1 GHz as well
<archetech> ah
<Tonymac32> the disk I/O is probably the main culprit
<archetech> I have ccache this time and fresh ubu 20 vs 18
<Tonymac32> I have it set with local ccache on the m.2
<archetech> I installed it and rebooted was that enough ?
<Tonymac32> ?
<archetech> ccache
<Tonymac32> the build script should handle that on it's own
<Tonymac32> so I don't know
<archetech> gah arch says export PATH="/usr/lib/ccache/bin/:$PATH"
<archetech> ill run ccache -s will say
<archetech> its wireless drvs that I could have totaly turned off grr
<Tonymac32> ok the postprocess script was there from Baylibre, I added a small fix in MAy, let me see if it works now
sunshavi has quit [Ping timeout: 260 seconds]
<Tonymac32> we don't have overlays on N2 yet?
<nekomancer[m]> ? there are overlays in folder in /boot
<Tonymac32> what are their file names? I'm not seeing a prefix for the G12
<Tonymac32> I don't want them confused with the GXL ones,m which I think is probably likely what's going on
<nekomancer[m]> meson-g12a, meson-g12b
<Tonymac32> hmmm, the "a" ones probably won't work
<nekomancer[m]> meson-g12b-odroid-n2.dtb
<Tonymac32> I mean the dtbo's
<nekomancer[m]> there was 5 or 6 overlays... sorry just now have no bootable N2 to look into
sunshavi has joined #armbian
<Tonymac32> ok, once I get past u-boot build issues I'll be able to see for myself
<Tonymac32> yeah, those look like the GX/GXL ones
<Tonymac32> I'll have to take a look at that
tehircz has quit [Ping timeout: 246 seconds]
<Tonymac32> it's only an issue if we try to eventually make a logical library of overlays, we have a similar problem in Rockchip and Rockchip64 where they are all called "rockchip"
<nekomancer[m]> and n2 still miss RTC and watchdog. despite dtb patches fly around from winter
<Tonymac32> I was watching some of those arguments
<Tonymac32> lol
<Tonymac32> I don't know who is actually the maintainer of this board here, maybe a bit of everyone so a bit of no one
tehircz has joined #armbian
<Tonymac32> Or maybe should be me and I didn't realize :(
<nekomancer[m]> chewitt have that. but they in her repo.
<Tonymac32> I wouldn't have added it with u-boot 2015 though. XD I learned that lesson from the C2 and Potato
<nekomancer[m]> are librem board on amlogic better than same cobfig rock64 or not?
<Tonymac32> the Potato?
<Tonymac32> It doesn't have any weird boot issues
<Tonymac32> quad-core A53 is basically exactly the same
tehircz has quit [Ping timeout: 246 seconds]
<Tonymac32> I think it might lack Arm crypto extensions somehow, or maybe that was only the K2/C2 S905
tehircz has joined #armbian
<Tonymac32> No USB3 on Meson GXL though
<Tonymac32> I'm getting scolded by my cat for ignoring it, lol
<Tonymac32> OK, u-boot built, good sign
<Tonymac32> hahaha
<Tonymac32> I forgot to suspend BOINC before I started this build, it disturbed the force
<archetech> Ill have to move this build vm to the ryzen 1600
<archetech> maybe... I dont plan to build alot
<Tonymac32> uptime: 31 days
<Tonymac32> lol
tehircz has quit [Ping timeout: 264 seconds]
chewitt_ has joined #armbian
tehircz has joined #armbian
chewitt has quit [Ping timeout: 260 seconds]
chewitt_ has quit [Client Quit]
WoC has joined #armbian
tehircz has quit [Ping timeout: 246 seconds]
tehircz has joined #armbian
<Tonymac32> lol a strange error from cleaning made me wipe out the proper boot image, rebuilding
<archetech> mines done gotta boot
<Tonymac32> there we go. Sometimes u-boot will compile even if it has a failure to execute the defconfig. which makes no sense, but there it is
<Tonymac32> the legacy is "odroidn2_defconfig", mainline is "odroid-n2_defconfig"
<archetech> wheres that picked
<Tonymac32> the board config
<archetech> if I choose current I get mainline I assume
<Tonymac32> but there's a buinch of other stuff I had to adjust first
<Tonymac32> that was just a snag I ran into
<archetech> I will redo the uboot after it boots
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
<archetech> it boots
<Tonymac32> hey, so did my mainline one
<Tonymac32> sweet
<archetech> no this was built 3 hrs ago
<archetech> no this was started 3 hrs ago
<Tonymac32> yeah, but I mean my test image booted
<Tonymac32> so I can work on rolling that out
<archetech> odroidn2 Kernel: 5.8.16-meson64 aarch64 bits: 64 Console: tty 0 Distro: Armbian GNU/Linux 11 (bullseye)
<archetech> Machine: Type: ARM Device System: Hardkernel ODROID-N2Plus
<archetech> Armbian 20.11.0-trunk Bullseye
<archetech> ah I can upgrade to that right
<Tonymac32> not yet, and probably would need a new image
<archetech> ill want it when you have 5.10rc1 that will have the new bifrost drv
<Tonymac32> lol maybe dev will have that at some point, but we'll be bumping Current to 5.9 first
<archetech> k
<Tonymac32> anyway, passing out time
<archetech> cya
<Tonymac32> @igorpec don't merge that PR yet, it would break legacy boot, the config name is different between legacy and mainline u-boot.
tehircz has quit [Ping timeout: 258 seconds]
<IgorPec> tonymac32: ok, won't - btw. just sent email about Honeycomb
tehircz has joined #armbian
<archetech> agetty[3063]: /dev/ttyS0: not a tty log spam evy 10 secs
tehircz has quit [Ping timeout: 258 seconds]
tehircz has joined #armbian
<archetech> systemctl stop serial-getty@ttyS0.service
<IgorPec> where do you see this?
<archetech> fresh n2 bullseye
<IgorPec> focal doesn't have it
<archetech> cat /proc/cmdline somethingin here is mentioned
<archetech> another issue elevator=noop
<archetech> kernel: Kernel parameter elevator= does not have any effect anymore.
<archetech> Please use sysfs to set IO scheduler for individual devices.
<IgorPec> that's lepotato
<IgorPec> the problem is that console is kernel dependend
<IgorPec> on legacy is configured different ... and perhaps somewhere we don't have correct info
<archetech> ya just gave a post for example
<archetech> i.e its not new or unique
<archetech> boards down now till tommrw
tehircz has quit [Ping timeout: 246 seconds]
archetech has quit [Quit: Konversation terminated!]
tehircz has joined #armbian
tehircz has quit [Ping timeout: 258 seconds]
tehircz has joined #armbian
archetech has joined #armbian
sassinak-work has quit [Ping timeout: 260 seconds]
sassinak-work has joined #armbian
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
Hokedli has joined #armbian
tehircz has quit [Ping timeout: 260 seconds]
tehircz has joined #armbian
tehircz has quit [Ping timeout: 240 seconds]
tehircz has joined #armbian
chewitt has joined #armbian
<ArmbianTwitter> @jerry_hopper (jerry hopper): The latest the Altis Times! https://t.co/VXnx9ld7HQ Thanks to @veterans_gaming @destprism @armbian #pakistanarmy #foss (9s ago)
<ArmbianTwitter> @armbian (armbian): @guidol70 Happy 1/2! :) (28s ago)
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
<ArmbianTwitter> @armbian (armbian): Check out audio bugfix update for @AmlogicPR powered Odroid N2 from #hardkernel enhanced by #armbian kernel 5.8.16 #wireguard #docker #debian #ubuntu #minimal #iot #infosec https://t.co/JuVtxG0CCQ https://t.co/v6GXdSoDfv (17s ago)
rneese_ has joined #armbian
scope has joined #armbian
scope has quit [Client Quit]
scope has joined #armbian
<lanefu> Tonymac32: fyi there's an option in github to set it as a draft pull requeset to keep it frorm beieng merged
scope has quit [Quit: Laundry time]
rneese_ has quit [Ping timeout: 240 seconds]
Hokedli has quit [Quit: Konversation terminated!]
tehircz has quit [Ping timeout: 258 seconds]
tehircz has joined #armbian
torv_ has joined #armbian
torv has quit [Ping timeout: 240 seconds]
<silver_hook> IgorPec, Werner: Ohai, sorry, somehow missed the ping. What did I miss?
<IgorPec> hey, how are you? nothing that critical ... can we ship Microsoft code together with image?
<silver_hook> Depends on the license, I’d say.
<silver_hook> There is MS FOSS out there as well (and increasingly so)
<IgorPec> so the q is if code from ms downloads is OK? are there two versions, which licence is cool except GPL? is it gpl
<silver_hook> Got a link? (PM also OK)
scope has joined #armbian
tehircz has quit [Ping timeout: 260 seconds]
<ArmbianTwitter> @DestroyedLolo (Destroyed Lolo 🎿🏔️🇨🇵🇪🇺🌍): So it's done : ✅My dashboard ported to #armbian ✅Board switched ⏰only remaining network routing ... This evening https://t.co/44ci3gY6Pq (20s ago)
tehircz has joined #armbian
rneese has joined #armbian
rneese has quit [Client Quit]
rneese has joined #armbian
tehircz has quit [Ping timeout: 260 seconds]
tehircz has joined #armbian
scope has quit [Quit: Laundry time]
<Werner> Good day
<stipa> xfce is very stable, going stron for 16 days without reboot running one VM...
<stipa> i'm amazed
tehircz has quit [Ping timeout: 264 seconds]
tehircz has joined #armbian
<ArmbianTwitter> @Nosluop56 (George P): Well that's my @thepine64 #PinebookPro sorted! Setting up @armbian just couldn't be any easier... From loading the image on SDcard to booting from emmc with root filesystem on nvme SSD takes just a few clicks! Perfect!👍👍 https://t.co/VfcYoYBbjM (26s ago)
DaRock has quit [Ping timeout: 256 seconds]
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
<ArmbianTwitter> @armbian (armbian): RT @Nosluop56: Well that's my @thepine64 #PinebookPro sorted! Setting up @armbian just couldn't be any easier... From loading the image… (29s ago)
<WoC> Speaking of, my quad os multi-boot works really good :) ArmBian is my 4th OS :) (pbp)
<IgorPec> not 1st? :( LOl
<WoC> No particular order, just the order in which they were installed :)
<Werner> ^^
<WoC> kernel 5.8.13 :)
<IgorPec> it has everything
<WoC> One and the same kernel in all 4, as it requires a kernel patch for the multi-boot
<IgorPec> but i don't have pbp so .. no idea how well or not is working
<WoC> Just waiting for the FreeBSD image for my 5th OS, as that is EFI and i need to figure out how to resize the FreeBSD fs's ;P
<WoC> and afaik, there is no way to add a efi in u-boot as a menu item
<buZz> uboot supports EFI?
<WoC> yup
<buZz> x86 only , it seems?
<WoC> and if one had been smart, i would have enabled iso9660 and udf, but like someone said; I may not be a smart man.... Life if like a box of chocolate ;)
<WoC> arm
<WoC> Linux Cursed 5.8.13-pinebookpro-arm64 #1 SMP Fri Oct 9 21:17:22 CDT 2020 aarch64 aarch64 aarch64 GNU/Linux
<WoC> (cortex-a53, cortex-a72
<WoC> panfrost still a pain in the pita... other than that, works great :)
<WoC> almost wrote painfrost ;P
<buZz> oh, TIL pinebook pro doesnt use allwinner :P
<buZz> another platform i can safely ignore \o
<WoC> RK3399
<WoC> heh
<WoC> Found one board with 24 cortex-a53 cores
<WoC> only $1250+ (USD)
<WoC> Up to 64 GiB ram though :)
<WoC> Probably will be more boards of that kind the next 2 years :)
<WoC> pcie x16 slot even :)
<buZz> WoC: did you look at the cavium chips yet?
<buZz> >40 cores armv8 iirc
<WoC> and a even bigger pricetag... ;P
<WoC> Aye, i think it's armv8-a or above
<WoC> just realized i forgot to install a web browser...
<WoC> would be nice with a full size atx board with plenty of card slots and a few M.2 for nvme along with 64+ GiB ram support :)
<WoC> This being my smallest install though... /dev/sda 15G 9.7G 4.1G 71% /
<WoC> Marvell Announces ThunderX3: 96 Cores & 384 Thread 3rd Gen Arm Server Processor
<WoC> ThunderX3: 2022 :)
<WoC> err
<WoC> ThunderX4: 2022 :)
<WoC> The ThunderX3 - 96 Cores and 384 Threads in Arm v8.3+ <-- Bet that would work really nice with distcc ;)
<IgorPec> we are looking to get one of those proper arm server for our needs
<IgorPec> currently we CC on dual xeon server
<WoC> I have a Quad Core Xeon now, distcc+crossdev
<buZz> ah right, marvell bought cavium?
<IgorPec> we need something much more powerful
<WoC> 32 GiB ram and tmpfs
<WoC> Power9+ ?
<IgorPec> recompilation of all kernel variants currently takes hours, almost half a day
<WoC> using tmpfs ?
<IgorPec> 128gb memory
<IgorPec> nothing else is on that server
<IgorPec> problem is that single core is not very fast
<WoC> That would fly, as long as the build env is on tmpfs
<IgorPec> it flies when it works
<IgorPec> most of the time is wasted on packing and preparing
<buZz> IgorPec: why is it done with single core?
<IgorPec> certain operations are
<buZz> i ment building the kernel
<IgorPec> that is multicore
<IgorPec> but i need packed kernel :)
<IgorPec> and i also need patched kernel
<buZz> ah
<buZz> hmhm
<IgorPec> then i need BSP package, need u-boot
<IgorPec> and between those operations, its lag
<WoC> and u-boot-bsp...
<Werner> How many images are currently built in parallel?
<IgorPec> i think i can fire 50
<IgorPec> but i can build images in parallel only when i have packages
<buZz> its manual?
<IgorPec> nope, that works perfectly
<IgorPec> if i select 100 images at once i will ran out of memory
<IgorPec> operaion is unpacking rootfs, installing packages, coying, packing, signing ... this is all multicore
<Werner> package creation is parallel too?
<IgorPec> no. that is not
<Werner> Why?
<IgorPec> phase 1 is to prepare packages ... then from there it go lightspeed
<IgorPec> well, because we alter sources
<IgorPec> sources are common, patches right before compilation, both for u-boot and kernel
<IgorPec> if i start this part in paralel it will crash since two kernels would patch sources differently ...
<Werner> If they use the same sources.
<Werner> What if there were multiple caches?
<IgorPec> yeah, but then you need to implement semaphore mehanism
<IgorPec> multiple caches would solve that, yes
<IgorPec> but not sure if its worth developing into that direction
<IgorPec> we will have less and less different sources
<IgorPec> and bootleneck is not kernel compilation
<IgorPec> everything else that leads there
<Werner> That is not the point. Some tasks while building packages are not using multicore which is wasted time. If we could in parallel idling would be reduced
<IgorPec> but that is sometimes hard to achieve
<Werner> True
<IgorPec> we already mady very far with all this
WoC has quit [Remote host closed the connection]
<IgorPec> btw. regarding Microsfot code
<IgorPec> if we want to ship it, we would need to build community version
<Werner> Are there issues with this? regarding plugins and such
<IgorPec> no idea about the plugins, but this is how the licence is interpreted
WoC has joined #armbian
<Werner> Hm should we invest time to build a custom vscode version to ship with images which increases their size and probably just a fraction of users will use? Those who want it know that they are just a single command ahead from using it at the moment.
<IgorPec> yeah, i agree its not worth the trouble
<Werner> What do you think about chwe's idea to integrate a redirector page for external links?
<IgorPec> yes, we could do that. Can we exclude our domain? But we need to buy some plugin?
<Werner> I was hoping to find something for free but stopped digging after a couple hundreds of topics. Anyway I am not sure if it is worth the money atm. As far from what I noticed there is one spamer every couple of days currently and most are catched from moderation.
<IgorPec> ok
<Werner> I added a symbol to any link that does not link to either armbian or github. They are whitelisted to say. Maybe this works out well.
<IgorPec> oh, that's already good enough IMO
<IgorPec> you can spot things easiery
<Werner> Probably need to add a couple more domains to the "whitelist"
<Werner> missed docs.armbian.com for example
drobo_00 has quit [Ping timeout: 260 seconds]
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
drobo_00 has joined #armbian
stipa has quit [Quit: WeeChat 2.3]
<ArmbianTwitter> @KahLoyx (kah@loyx): @Nosluop56 @thepine64 @armbian Is it battery killing to run root filesystem on a nvme 😄? (18s ago)
<WoC> maybe, pretty much depends on which and in what ?
<WoC> not to mention if it's a heavy load web server...
stipa has joined #armbian
<ArmbianTwitter> @Nosluop56 (George P): @KahLoyx @thepine64 @armbian Too early for me to tell really. I've configured the nvme to minimise power consumption and it's running really well 👍 (23s ago)
stipa has quit [Quit: WeeChat 2.3]
stipa has joined #armbian
archetech has quit [Quit: Leaving]
willmore has quit [Ping timeout: 240 seconds]
<ArmbianTwitter> @Madiator2011 (Madiator2011 Gaming): @Nosluop56 @thepine64 @armbian Will need test it! (11s ago)
<IgorPec> done
<Werner> Awesome
tehircz has quit [Ping timeout: 240 seconds]
tehircz has joined #armbian
archetech has joined #armbian
stipa has quit [Quit: WeeChat 2.3]
stipa has joined #armbian
willmore has joined #armbian
willmore has quit [Ping timeout: 240 seconds]
willmore has joined #armbian
tehircz has quit [Ping timeout: 264 seconds]
tehircz has joined #armbian
<Tonymac32> is the thunder Xn still happening? I thought they had shut the door on general-purpose SoC's like that
<Tonymac32> "Based on highly efficient full custom processor cores designed in 28nm processor technology"
<Tonymac32> heh
<WoC> No idea, but sure would be nice with a full sized ThunderX3 or X4 Mobo, with plenty of ports/slots
<lanefu> I'd assume they're planning on doublign down on anything datacenter oriented
<lanefu> amazon opening the ARM floodgatse for the hyprescalers
Elpaulo has quit [Quit: Elpaulo]
<WoC> would be neat with the cavium mobo and a nv card for cuda on arm :)
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
<archetech> my 04 blzer has apinched nerve in the steering colomn
<archetech> cant move shifter
<Tonymac32> lol a binched nerve?
<Tonymac32> pinched*
<Tonymac32> arg
digitalman1983 has joined #armbian
<digitalman1983> hi all
<digitalman1983> I'm currently working with a trunk build for the A64 using the legacy kernel (5.4). While this has greatly improved boot repeatability, it seems my USB ports are broken
<digitalman1983> is this a known issue?
<[TheBug]> thats a pretty open question
<[TheBug]> what A64 board?
<[TheBug]> do you have the correct DTB settings for the ports set?
<[TheBug]> if you can give more information maybe we can try to better answer your question
<[TheBug]> also maybe which specific image?
<lanefu> archetech: i believe it.. a ton of electrical stuff was terminatd off the ignition switch...
<lanefu> 04? that's like the last year isn't it?
<archetech> 2004
<archetech> the truck that pays ME to IT
<archetech> the truck that pays ME to drive IT
<lanefu> lol
<lanefu> i had a 99, heads cracked
<archetech> this has only 90k
<lanefu> got new motor... fuel pump went and had to change it in the parking lot
<lanefu> wow
<lanefu> time machine
<lanefu> 2door?
<archetech> yup red ls
<lanefu> nice man... they were cool... ihada an 86, 93, and a 99 s10blazer
<digitalman1983> thebug: this is an Olimex Lime A64
tehircz has quit [Ping timeout: 240 seconds]
<digitalman1983> I am using the Armbian build scripts to create the image
<[TheBug]> digitalman1983: yeah see I would have never have guessed that information at all based on what you said, I would have assumed you were using a prebuilt image and having issues
tehircz has joined #armbian
<[TheBug]> but it would sound like you don't have the correct DTB or parts are missing from it which are needed for those USB ports to be active
<[TheBug]> maybe worth grabbing a prebuilt image if one exists and comparing the DTB you are using?
<digitalman1983> sorry, thought "trunk build" was specific to self built
<[TheBug]> well we have a lot of internal builds flying around for testing that a trunk builds so.. yeah thats still a bit confusing sometimes ;p
<lanefu> [TheBug]: digitalman1983 has been working on thee Lime stuff for a while... Igor promoted back to csc for digitalman1983
<[TheBug]> ahh
<[TheBug]> I am out of the loop it seems, my apologies ;X
<digitalman1983> it's all good, I appreciate every last bit of help I get in here!
<digitalman1983> I'll take a look at the device tree tonight and see if it's missing anything that the released build has
<digitalman1983> I do know that u-boot recognizes my hub and devices....as soon as the kernel starts, the lights go out and USB devices cease to function
<[TheBug]> and when in OS you don't see anything under lsusb?
<[TheBug]> it would seem you are not only / first to see similar issue in some case
<lanefu> digitalman1983: add this to your customize_image.sh script sed -i 's/verbosity.*$/verbosity=7/' /boot/armbianEnv.txt
<lanefu> or just put verbosity=7 in /boot/armbianEnv.txt directly
<lanefu> that should at least show you if its kernel related or not
<[TheBug]> digitalman1983: out of curiousity, since I read something that discussed it, have you tried physically disconnecting / reconnecting the devices after boot, one things I read mentioned it worked after re-plug but not sure if thats same issue..
<stipa> does anyone have ARM PC? Like using it all the time instead X86 stuff?
rneese has quit [Quit: Leaving]
<digitalman1983> thanks guys, just leaving work now, will give these few things a try when I get home....thanks a million!
digitalman1983 has quit [Remote host closed the connection]
<buZz> stipa: yeah my laptop is a former chromebook
<lanefu> stipa: i use an N2 and a Pinebook pro for light desktop stuff
<buZz> i do have a x86 laptop aswell, but i only carry the arm laptop around with me
<lanefu> satisfies all my needs for weekend hacking on armbian etc
<buZz> exynos 5800, 4gb ram, 10" 1080p
<buZz> arm laptop is lighter, thinner, has longer battery life, and no fans :P
<Tonymac32> archetech I have a coworker with a late 90's blazer at 290k
<Tonymac32> but yeah, if the brake switch interlock hicky is broken it won't let the shifter move
<Tonymac32> I can't remember if those are electric or mechanical on those
tehircz has quit [Ping timeout: 265 seconds]
<Tonymac32> there might be an interlock on the key switch as well actually
<stipa> buZz: lanefu cool, i see on some Youtube videos that desktop PC arm machines also don't have fans.
<stipa> that's sick
tehircz has joined #armbian
<lanefu> stipa: like if you just do terminal stuff and a little youtube they're quite usable
<Tonymac32> stipa my Tinkerboard served as my desktop for a few weeks between machine repairs
<Tonymac32> The N2 I imagine is quite serviceable
<lanefu> yeah N2 feels super natural when using it
<lanefu> like it felt fastre than that cherry trail rock pi X
<Tonymac32> 4 big cores are nice
<Tonymac32> rock Pi X is made of leftovers isn't it?
<stipa> i see you guys are rocking it hard.
<Tonymac32> ;-)
<lanefu> ha rock Pi X competes with Robot Hearts
<Tonymac32> I also have a PBP, but it hasn't entered active duty just yet
<lanefu> stipa: i'm commited to beeing a nerd
<Tonymac32> I need to boot Armbian on the PineTab
<Tonymac32> :)
<Tonymac32> should be straightforward, same as the pinebook non-pro
<lanefu> yeah
<Tonymac32> the foldup case is nice on it
* Tonymac32 actually wants to stick WebOS on it
<Tonymac32> :D
<lanefu> i wondre what perf difference is betweeen A64 and H5
<Tonymac32> YUGE
<Tonymac32> lol
<Tonymac32> I honestly don't know, never looked
<Tonymac32> My build system wants to know why I'm compiling a 4.x kernel
<lanefu> dude my buddy is using my garage to mess with a '69 volvo.... 1" oil drain plug
<Tonymac32> hahahahaha sheeeesh
tehircz has quit [Ping timeout: 264 seconds]
<Tonymac32> gotta make sure that 50-weight can come out
<lanefu> naturally i had to give him my 1/2" braeker to turn it
<lanefu> hahahaha
<lanefu> yaeh its like a B20 engine.. not sure what thee history is on that, but at least seems straight forwrard
<lanefu> and it already has a weber carb conversion
tehircz has joined #armbian
<lanefu> last time i changed oil in my jeep, ij ust went with Rotella T lol
<Tonymac32> I removed the bed from my Ranger in my driveway, snapped off 1 bed bolt, ruined the other 5 anyway. Had to use my breaker bar with a socket and extension over the end to turn them
<stipa> did you guys have experience with those arm server machines with 64 cores or so?
<Tonymac32> friggin jeep
<Tonymac32> stipa no, I'm trying to get my hands on a 16 core NXP based one
<Tonymac32> still around $1000 though
<lanefu> stipa: i've ran a few AWS graviton VMs they're preetty decent
<lanefu> need to try graviton2
<Tonymac32> I need to win the lottery
<Tonymac32> XD
<lanefu> the big thing about the giant arm servers other than core count, is memory IO
<Tonymac32> right
<lanefu> its generally superior
drobo_00 has quit [Remote host closed the connection]
<stipa> it's pretty sick
<lanefu> general single thread perf is samee or less than 2.3ghz xeons
drobo_00 has joined #armbian
<Tonymac32> that NXP chip is supposed to be a switch, that's why it has 4x SFP+
<lanefu> I did benchmark a 4core graviton
<stipa> Tonymac32: which board has that 16core NXP?
<Tonymac32> and it's mini-itx
<lanefu> would be interesting to get java benchmarks on those
<Tonymac32> ew
<Tonymac32> wth
<Tonymac32> gross
<Tonymac32> stop
<Tonymac32> XD
<lanefu> like do you think it's cpu cache is light compared to a server CPU
<stipa> oh my
<lanefu> if its meant for forwarding traffic, probably doesnt need a lot of ondie cache
<Tonymac32> yeah hang on, NXP = documentation
* lanefu could be full of shit
<stipa> Tonymac32: so everything is in Soc?
<Tonymac32> needs RAM
<stipa> i guuess that's why it's so big
<Tonymac32> looks like it has 1 MB L2 per pair of cores
<Tonymac32> which is a bit light
<lanefu> yeah
<Tonymac32> 8 MB L3 I guess you'd call it
<Tonymac32> notice dual-channel ECC DDR4
<Tonymac32> right
<Tonymac32> my game machine
<stipa> They are marketing it to IoT market
<Tonymac32> XD
<Tonymac32> not really a great processor for that
<Tonymac32> but
<lanefu> 100gigabit IOT
<Tonymac32> dude you need high res video uncompressed streams on the edge
<stipa> oh wow, look at that rack cases for it...
drobo_00 has quit [Remote host closed the connection]
drobo_00 has joined #armbian
<stipa> my bad, it's for some other stuff
<Tonymac32> lanefu 768 GB RAM
<Tonymac32> XD
<Tonymac32> they're both LGA2011-3
<Tonymac32> lol
<lanefu> lane@billroyall:~/GIT/build$ free
<lanefu> total used free shared buff/cache available
<lanefu> Mem: 264025940 32448904 52258912 14567876 179318124 215292920
<lanefu> Swap: 0 0 0
<lanefu> im barely getting by
<Tonymac32> I'm running 64 GB in my main machine, 16 in the one I actually build with
<lanefu> ha this is eeasier
<stipa> hmmm, do you have an -h switch maybe?
<lanefu> lane@billroyall:~/GIT/build$ free -h
<lanefu> total used free shared buff/cache available
<lanefu> Mem: 251G 30G 49G 13G 171G 205G
<lanefu> Swap: 0B 0B 0B
<stipa> oh my
<Tonymac32> stipa lanefu LOL
<stipa> is that 256GB of ram?
<lanefu> yeah it absurd
<Tonymac32> that's what that looks like
<lanefu> i've got like a massive tmpfs volume and gave rneese a VM with a decent amount of ram and i still have that left
<Tonymac32> 171GB of cache
<Tonymac32> rofl
<lanefu> that server had more ram than storage for a while
<Tonymac32> I'm only using about 16 of the 64 GB I have available with about 110 firefox tabs and rosetta running
<lanefu> i had 2 64gig Sata doms on it and that was it
<Tonymac32> hahaha
nologin has quit [Ping timeout: 272 seconds]
<stipa> sick stuff
<stipa> strange how it doesn't crash...
<stipa> running all that stuff
<Tonymac32> My Helios64 shipped, so I should have a home for my 20 TB of drives
<Tonymac32> deciding on filesystem
<stipa> oh my
nologin has joined #armbian
<stipa> how do you connect that to a pc?
<Tonymac32> it is an RK3399 computer
<Tonymac32> so it's a NAS/Server
<stipa> but
<Tonymac32> I have an XU4 running that job at the moment
<archetech> wonder how neko likes his
<stipa> how fast can you send data to another machine?
<Tonymac32> I'm being called away though, I'll BBL
<stipa> i see, Up to 2.5Gb Ethernet
<stipa> 312 MBs
<[TheBug]> RockPi 4c is pretty good also as NAS with a PCIe adapter and a 8 port sata card
<[TheBug]> so Helios64 is gonna have to be pretty sweet
<[TheBug]> what I really want is the version they are discussing to come with some ECC type ram
<[TheBug]> that would be sweet
<stipa> hmmm
<lanefu> Helios64 is supposed to support a DAS mode via USB.. so in theory 5gigabit IO to a PC
<stipa> which sata adapter?
<stipa> lanefu: i see, that's fine speed.
<[TheBug]> lanefu: I have some doubts there, are the usb 3.0 and pcie lanes shared?
<[TheBug]> I mean 2.5Gbs sounds possible and realistic
<[TheBug]> would be cool if you could get full bw though in t hat way
<stipa> it looks like it could
<stipa> looks fancy
<stipa> it's cool how network is slowly becoming a bus
<stipa> for peripherals
<lanefu> [TheBug]: yaeh you'd have to ask the kobol guys or look at the schematics.. no idea
<lanefu> stipa: so true... like i have an HDhomerun.. perfect example magic box just sitting on my network
<stipa> yeah, it's hard to explain
<stipa> like
<stipa> it's one big computer
<stipa> the network
<stipa> and you have multiple access points
<stipa> be it smartphone, desktop pc, laptop... a watch
<stipa> that connect to it
<stipa> to the resources
<stipa> and share them
<stipa> betweene themselves
<lanefu> lol there used to be TV commercial for IBM or nortel that said "The network is the computer"
<stipa> right, it always was but we didin't see it because of the slow speeds of the network
<stipa> sick
drobo_00 has quit [Ping timeout: 244 seconds]
<stipa> a 2.5Gbs starnsfer is so fast that you can boot a ps from it on another machine on the network
<stipa> a pc*
<stipa> not just one, at least three computer on the network could have oses on that NAS disk without a problem...
drobo_00 has joined #armbian
<stipa> practically, you could have a machines without diskis in them
<stipa> that use nas as their main disk
<stipa> but is it practical i dcan't tell
<stipa> probably not
epsilonKNOT has quit [Quit: ZNC - https://znc.in]
ddurst has quit [Ping timeout: 246 seconds]
tehircz has quit [Ping timeout: 256 seconds]
tehircz has joined #armbian
ddurst has joined #armbian
<nekomancer[m]> what rootfs somebody yesterday ask for?
epsilonKNOT has joined #armbian
tehircz has quit [Ping timeout: 272 seconds]
tehircz has joined #armbian
<[TheBug]> stipa: you can already run boards without disk via TFTP/nfs/iscsi/NBD
<[TheBug]> so it isn't a 'could' it is a 'you can already do that'
<stipa> [TheBug]: yeah thanks, i google it, it's called "Diskless system".
<stipa> Diskless node
<archetech> nekomancer[m] ya havent said how the new nas is
<lanefu> i use to run my garage computer off of PXE boot, and use ubuntu ovr NFS
<lanefu> it was pretty cool
<lanefu> even cooler i could boot 2 or 3 machinees off the same pxe setup and it was fine because they used tmpfs for /var/run etc
<archetech> i ran a bunch od vm's on proxmox over nfsv3
<archetech> of
<archetech> diskless pmox
<archetech> I useda small pmox boot hdd though
<archetech> vm's booted fr nfs
<stipa> lanefu: archetech sick
<nekomancer[m]> <Tonymac32 "171GB of cache"> "build for ZFS"
<stipa> the only downside to that is that someone needs to carry nas together with machine that uses it if leaving the network
<archetech> its a production setup who would move it
<archetech> kids
<nekomancer[m]> <[TheBug] "what I really want is the versio"> no arm chips with ECC support inside (not only on ram bus, but caches, registers etc)
<stipa> the next cool thing would be to boot oses from the internet, in that case you could be wherever you have access to the internet.
<stipa> heck, you could boot it from your home over the internet
<stipa> from the same nas
<stipa> sick
<stipa> but i guess the latency would be a problem
<stipa> archetech: right
<stipa> maybe if you have diskless laptop
<[TheBug]> nekomancer[m]: go read, they are looking into releasing a special memory chip with support -- Helios said the following:
<[TheBug]> "We will offer an ECC option. The Rockchip RK3399 SoC itself doesn’t have a memory controller that has ECC feature, however we are currently working with a SDRAM vendor that now offer built-in ECC feature inside the SDRAM directly. It’s not impossible it will be available on day one"
<nekomancer[m]> <archetech "nekomancer ya havent said how "> still not get HDD for them. want 2×14 TB, but costs and SMR scandal... Hard to decide
<archetech> I C
<[TheBug]> stipa: you can boot from the internet too via PXE
<[TheBug]> it supports pulling boot media from http url
<stipa> [TheBug]: have you tried that?
<[TheBug]> yes
<stipa> is it usable?
<[TheBug]> well you are talking booting not running filesystem from it
<stipa> how does it feel?
<stipa> oh no writing to it?
<nekomancer[m]> <[TheBug] ""We will offer an ECC option. Th"> seems kind of hentai for me.
<[TheBug]> the boot process is completed before anything loads, PXE downloads media to memory and then runs it, it is no different than normal
<[TheBug]> booting and storage a different things
<[TheBug]> though you could also mount NFS over the internat or iscsi if your brave
<stipa> i see
<[TheBug]> I am sure would be slow if you have a slow connection
<[TheBug]> nothing restricts you doing it though
<stipa> 5G could make it usable maybe
<stipa> like having a "hard disk" in a cloud thing.
<stipa> with os and everything
<[TheBug]> I mean if you are on a gigabit connection and have things with-in 20-30ms you probably wouldn't really notice much for small tasks
<[TheBug]> but really you would run most things in ramfs like a livecd
<[TheBug]> so standard tasks would probably be fine
<stipa> i see