#devs_os-archived

1 messages · Page 3 of 1

west hinge
#

Pascal, could you please have a look? Both to the PR, and preferably to the issue...

#

PR's done, thanks @dusk phoenix . Would still appreciate comments on the issue itself 😉

west hinge
#

ok, next question on the matter: using the partition install, I want to change SSD's. Ordered a new, bigger and faster, and (hope) more efficient one 😉 how can I copy/move the content of my current SSD to the new one?

west hinge
#

the official partition install is quite impossible to reboot the host. A powered hub has back-power preventing the Pi, a tweaked USB cord (no red power wire) doesnt reboot either

#

apparently a 3.5+A PS does work (Thanasis) but these are not obvious in all regions. Believe a list of positively tested Hubs/PS's/usb cables should be provided in the install method, or people might burn their PI's...

upper nebula
#

Hello devs

#

I am trying to boot odroid c2 image on khadas vim1 yes I am using short cut method to test my theory

dusk phoenix
upper nebula
#

Flashed c2 image on sd card.
Flashed Vim1 mainline uboot
Replaced dtb

#

Ok

#

please see if you can find the mistake here
I am trying to look for what I missed

#

ok I will edit the boot.scr can you point me to the boot.txt on the git will save some of my time

dusk phoenix
#

That might be a problem:

Unknown command 'setexpr' - try 'help'
upper nebula
dusk phoenix
#

Yeah

upper nebula
#

looking at the load mmc io is switched between c2 and vim1 uboot

dusk phoenix
#

You built your own U-Boot right? I think I'd try to add the HAOS sauce to it so the default script works

#

We use devnum which is provided by U-Boot, so it should work on any mmc device afaik

upper nebula
#

idk why setexp is not in the mainline uboot

dusk phoenix
dusk phoenix
#

fileenv is an extra patch though

upper nebula
#

ok I will look for setexp patch

#

well setexp is part of uboot maybe just need to enable it as you said

#

If I could get it to work with vim1 then it will work with vim3 also using n2 image

rare sable
#

the manufacture need send us 1-2 samples for each supported device that we can test with (new releases ect..)

#

I reach out khadas a year before and they had no interests to run HAOS

latent wigeon
#

new profile pic 👍

rare sable
upper nebula
#

tried compiling uboot for vim2 and it fails with some error using the khadas tool.

#

fails to compile dtc

upper nebula
#

Thanks for the replies devs. I gave it multiple tries but the uboot which I build doesnt build. Maybe the deps are not supported for the dtc the tool was using

#

I will try again tomorrow as I have asked the maintainer to add setexp and fileenv support to his builds.
Also will discuss about Khadas devices for hass

#

wait I have $100 coupon from Khadas store. @rare sable or @latent wigeon if you want to use it to buy khadas devices then I am willing to let you use it.
I have this since 6 months now as win yearly forum contributor price.

#

Can get 2 khadas VIm1

dusk phoenix
#

I am personally more interested in Vim3 🙂

upper nebula
#

That will be 1 device only

#

without the case maybe I can convince them to send case for free hahaha

dusk phoenix
#

Do you have a Vim3?

upper nebula
#

I have yes

#

VIm1 , vim2, vim3 Edge-V

#

all donation except vim1 which I purchase to begin my aarch64 journey 😉

#

I have dropped a message to Khadas manager. So either its 2 x Vim1 frm my coupon or 1 x vim3 frm my coupon and Khadas send one by themselve.
Lets see what response I get 😉

#

oh dear Vim3 basic is $99 i thought it was VIm3 pro. which have 4gb ram. anyways lets see.

#

Ok I will stop talking to myself. Off to sleep Bye guys see you again tomorrow

dusk phoenix
#

Hehe sorry, a bit busy right now 🙂

#

Good night!

dusk phoenix
upper nebula
#

Are you the only arm dev here ?

#

Ok then laters

dusk phoenix
#

pvizeli did most of the work in the past, now its mostly me

upper nebula
#

Hello

upper nebula
#

@dusk phoenix and @rare sable

Is 2 devices needed to add official support ?

#

Khadas team will let me know their discussion today hopefully.

dusk phoenix
#

Sounds good!

upper nebula
#

If i dont get a reply frm Khadas by sunday then I will go ahead and order Vim3 for you @dusk phoenix While I will build vim1 support myself

dusk phoenix
#

@upper nebula that is very kind. Let's hope that Khadas response

upper nebula
upper nebula
upper nebula
#

Looks good.
Khadas Vim1 setting up supervisor.

upper nebula
#

All good works so much better than rpi3B+

#

Next need to prepare an image out of my sd card to flash it on the emmc 😉

dusk phoenix
upper nebula
#

for rootfs not found it was wrong PARTUUID in boot.scr

dusk phoenix
upper nebula
#

how does the hass os tool build images ? dockers

#

which host is needed ?

dusk phoenix
#

The image building process is rather custom scripts

upper nebula
dusk phoenix
#

To build I use this command usually:

./scripts/enter.sh make O=output_odroid_c4 odroid_c4

(with that I have a separate output dir for each board)

#

odroid_c4 -> ./buildroot-external/configs/odroid_c4_defconfig

#

Ideally you want to create a custom khadas_vim1_defconfig + add what is necessary in ./buildroot-external/board/khadas/vim1/

#

Since its the same SoC I guess you shoudl be able to just copy ./buildroot-external/board/hardkernel/odroid-c2/hassos-hook.sh, adjust the device tree which should copied and then the image should be generated with all the right firmware etc.

upper nebula
#

@dusk phoenix
PARTUUID="4861cc6c-06" on C2 Image while its uboot-boot.ush have PARTUUID=48617373-06
This is on official image

dusk phoenix
#

That should not be the case. We set the UUID to 48617373-06 in buildroot-external/scripts/hdd-image.sh

upper nebula
#

Need minimum 16gb coz i used 16gb card as test image which I write back as img

upper nebula
#

@dusk phoenix does uboot ever gets updated on odroid or rpi devices?

upper nebula
dusk phoenix
dusk phoenix
#

Is there a specific reason you chose U-Boot 2021.07-rc4 (e.g. is it known to be required or might 2021.04 work too?)

#

Reason I ask, U-Boot sometimes causes a bit of trouble to update, so I usually don't jump on every release. And 2021.07 might be one of the skipped ones

dusk phoenix
upper nebula
upper nebula
dusk phoenix
upper nebula
upper nebula
#

Damn Khadas Vim1 is like double the speed of RPI3B+ when loading grafana

#

even though both have exact same soc just 100mhz overclockedon each core

#

when does the kernel get updated ? or does it stays on lts as long as it can ?

#

ok off to bed see you later

static tinsel
#

Hey @upper nebula I am not an "official" developer but I did implement the Odroid-C4 support. I was fortunate in that it's very similar to the Odroid-N2, so I didn't have to do much. I sometimes found it helpful to refer to Armbian source to see how various bits are put together...

upper nebula
#

Thanks

upper nebula
#

I am looking at hardkernel-boot pkg and will follow something similar and make khadas-boot pkg

bleak tide
#

Hi haos devs and thanks to @rare sable for inviting me as a haos dev myself. What an honor! However, before contributing more, what is the exact procedure for accepting/merging pull requests? Is there some documentation on the accepted development process (how and where - which branch? - to merge changes, when is review by another haos dev necessary and when not?). So you see I do have some questions here and hope perhaps @dusk phoenix or @rare sable can provide me some fist hand information to better know your development process...

dusk phoenix
#

@bleak tide since there aren't that many devs on OS, we don't have a very firm development process. Pascal for the most part transferred "ownership" to me. Typically, I test my changes at least on one platform locally and then create a pull request. Sometimes the PR is to just have it double checked by the CI pipeline, but depending on the change (especially if a bit more involved/or even controversial) I ask @rare sable to double check/give feedback on it. If I don't plan on doing a dev/release build, I usually let the PR sit there in case somebody else feels like reviewing it (which unfortunately doesn't happen all that often 😉 ).

All merge request should go to dev branch. Whatever is labeled rel-6 (or whatever is the current stable release) I'll backport before releasing the next stable. In this release cycle I built the rc's from dev branch, but I might do it differently with the next release cycle (I am planning to branch off a rc branch when releasing the first rc so development can continue while stabilizing on the rc branch). Since there are very few contributions currently, I'll expect developers to create PR's to the dev branch only and tag them if they feel it should be backported. I'll review/approve from all contributors.

We have two Github build pipelines: .github/workflows/dev.yml which requires manual trigger via Github actions, builds get pushed to https://os-builds.home-assistant.io/ and the dev channel if successful. Then there is .github/workflows/release.yml which builds on Github release and builds get pushed to the beta channel. From there we promote it to stable manually, if deemed stable worthy :).

bleak tide
#

@dusk phoenix Thanks for the details. This is highly appreciated. So, did I get it right? All PR should be merged to the dev branch regardless if they are labeled with "rel-6"? So this is just some information for you to don't forget to backport them to the "rel-6" branch when you are preparing a new stable release, right? So I should not directly commit/merge to the rel-6 branch if I want some change to be merged for the 6.x line of HAos, right? And I will of course leave it to you to prepare new stable releases but eventually let the dev workflow to generate new development builds.

dusk phoenix
#

Yes, that is correct, generally to dev branch and I use the label rel-6 as backport marker. There are situation where dev diverged/moved on so that a fix only really applies for the rel-6 branch, in that case a PR to rel-6 is ok.

#

Stable releases is a bit of work since I usually check it on various boards, so it depends a bit on schedule on my side when to prepare such a release. Development build I pretty much can build whenever.

upper nebula
dusk phoenix
#

On a side note, the VIM3 may soon be getting official support from Home Assistant OS as well, and we want to thank the open source community for their efforts!

#

I see... I don't remember promising anything 😄 We'll see

upper nebula
#

Yea but I dont think it will take much time but anyways enjoy ur vacation first 😉

upper nebula
dusk phoenix
#

Interesting, seems to be their first one

#

A bit more on the lower end, but nice price

sacred spear
#

is there any reason the home assistant host os on rasppi would be sending constant arp requests?

#

31148 916.611785 Raspberr_ce:b3:6d Broadcast ARP 60 Who has 192.168.86.64? Tell 192.168.86.65
31157 917.530036 Raspberr_ce:b3:6d Broadcast ARP 60 Who has 192.168.86.64? Tell 192.168.86.65
31158 917.530036 Raspberr_ce:b3:6d Broadcast ARP 60 Who has 192.168.86.64? Tell 192.168.86.65
31163 918.659450 Raspberr_ce:b3:6d Broadcast ARP 60 Who has 192.168.86.64? Tell 192.168.86.65
31164 918.659450 Raspberr_ce:b3:6d Broadcast ARP 60 Who has 192.168.86.64? Tell 192.168.86.65

#

seems to go away when the host os is restarting
and it's usually for the address right below the HA server address

dusk phoenix
#

The reason is probably that some component wants to send something to 192.168.86.64. I guess the question is what tries to send a IP packet to 192.168.86.64 (which in turn leads to said ARP request).

upper nebula
dusk phoenix
#

Yeah its even a bit faster probably, as its the S905Y2 clocked at 1.8GHz. But the VIM1 starts at 2GB, whereas this starts at 512MB. Definitly not where you want to run HA on 🙂

#

But yeah, the 4GB LPDDR4 with 16GB eMMC, WiFi5/BT5 @45$ is really a nice price!

upper nebula
#

Yes with a decent heatsink and case it would be good. Just Ethernet is missing though.

upper nebula
#

Any progress on Vim1 ?

woven crypt
#

I have installed the HomeAssitant OS inside proxmox. Everything works. Now I connected a Ralink Wifi dongle to my host and add this to the HAOS VM. nmcli does not detect the wifi dongle. Am I right that there are no Ralink drivers in the ova build ?

gleaming forum
dusk phoenix
#

planning to add them.

upper nebula
#

Hello Devs

serene reef
#

Hope this is not the wrong place to ask, but I imagine what I want to ask doesn't yet exist...
Problem: installation of hassio on RPi is too cumbersome and time consuming considering how much interaction is required by the installer.
Question : similar to the wpa_supplicant.conf file that is put in the boot partition of a RPi's SD card prior to installation, could a similar text file be inserted into the Hassio image boot partition to activate all the config and installation steps I would do right up to the point of being ready to add Iot devices?
WHY? I experiment with my setup quite a bit and seem to mess up my setup. I would imagine it using a package manager like YARN to install all the integrations, usb adapters and add-ons, docker images to be ready to go on MY network so I just need to add devices.
This could all be provisioned using something like the reprap configurator that would output the text/yaml/json files to be added to the SD card image.
Optionally, a repository of "configurations" for specific versions of parts could exist to duplicate a popular combination of integrations and hardware parts that are working together (known as parts working together). In my mind, it would be similar to docker hub for Home Assistant where I just add my network details and secrets file.
This would save many users hours or even days it takes to have a working starting place, which could dramatically speed up user adoption of the platform.

hollow lintel
serene reef
#

Thank you. I will ask there, but FYI, I am not talking about backups, I am talking about automated installation

#

... according to known, working recipes of of parts

rare sable
#

Devs channel are for people they want develop something

fierce zealot
#

Hello, new here, used the HAOS KVM qcow2 image to provision a VM, was absolutely shocked to see it drop me to a HA CLI that i could change files from or reboot the VM from, is that something normal? Removing the video and serial components are one thing, but really no login prompt ? Should i just do a supervised install instead ?

latent wigeon
#

This is a developer channel

#

you could always log in as root, so this just eliminates the need to do that

#

you can install whatever you want - this isn't a security issue

fierce zealot
#

@latent wigeon that's the problem, there is no login prompt, you just get the root shell

latent wigeon
#

What’s the problem with that?

fierce zealot
#

anyone with half a brain in your network can just destroy your install

#

if you don't remove the spice interface anything on your host can just about do that

#

was just curious if i did something wrong in the install that made it drop to an un-secure state

latent wigeon
#

Not really. They either have to have physical access to the device or network access to the machine, which you can protect

#

No, that’s by design

#

In the future, please use your install method channel, not the dev channels

fierce zealot
#

roger that, thanks

rare sable
#

Physical yes, network no

latent wigeon
#

I was thinking network to a machine running a VM, which should be protected

queen trench
#

I ran in the past the HAiOS in VM and it went smooth, now i change 🖥️ and wanted to do the same but now i have an issue. The VM runs well until I have to call it in the browser I'm not able to open it anymore. Who can help me

hollow lintel
upper nebula
#

Home assistant cli is failing to start on boot in a test image, Agner seems busy so I am asking here

hushed ravine
#

Was it more discussion of allowing more override for things like files, mounts, add nfs4 and samba file systems etc for a bit more user controlled flexibility of loading custom code at boot on home assistant os? I can’t see it compromise os if the user can make a mount point to mount a nfs share in for example media.

rare sable
#

We have an open RFC for implementing mounting into our eco system. Only that way will be allowed. Once a developer will implement that, we have a solution

#

All other workaround will not get merged

#

We doing it right or nothing

#

If you want customized OS, use debian and supervised

hushed ravine
#

Yeah I think I raised that rfc. Im happy to hear its in the pipeline. I don’t think customised os is the correct wording, but configured os. Im all for simplicity and keep it stable so I see the problem in opening it up. I hope that rfc get traction at one point. Good job on the os 🙂

upper nebula
#

Hello devs. Good to see the new release.

Was expecting vim3 but it's still not there yet.

rare sable
#

Vim3 come with rel-7 around October

#

We do not backport board support

dusk phoenix
ashen vine
#

Hi all, any hint on how to install a dev release of OS?
I'm struggling with Raspberry stops responding after few hours if I install anything newer than OS 6.0 and I'd like to try the 7.0...

hollow lintel
ashen vine
leaden lance
#

all - trying to troubleshoot the RPi4 crashing issue (#1119). trying to run journalctl and I get a response "unknown command". any idea what I'm doing wrong?

#

I should clarify that this is via ssh on the debug port 22222 login as root

rare sable
#

maybe the PATH is wrong?

leaden lance
dusk phoenix
#

Definitly works here:

$ ssh -p 22222 root@ha-shelf1-rpi4.lan
Welcome to Home Assistant OS.

Use `ha` to access the Home Assistant CLI.
# whereis journalctl
journalctl: /usr/bin/journalctl
#

Ah so you have HAOS 5.x? Use login to access the shell

leaden lance
#

^^fixed it. thank you sir

dusk phoenix
#

you're welcome

upper nebula
#

Hello Devs, hows everything ?

upper nebula
#

https://pastebin.com/WyPL6cCU

I have build image for Khadas vim1 using agners hassos branch

Supervisor cannot be started as the upstream source doesnt have it known

#

where is this defined ?
No version found for ghcr.io/home-assistant/khadas-vim1-homeassistant
We need to use odroid-c2 as it is present upstream

upper nebula
#

Can someone advice me on this ? Agner is travelling so he might not be able to respond.

#

In last nights image I was able to get hass_cli up as compared to previous dev images.
Now supervisor seems to be configured to use khadas-vim1 where it should use odroid-c2 instead

hollow lintel
upper nebula
#

great thanks, Do I have to rebuild the whole thing again or can I just rebuild the supervisor docker ?

hollow lintel
#

I have no clue how that works 🙈

upper nebula
#

No problem I have removed docker_cli and rebuild it

#

now supervisor is up yeayyy

#

next ethernet is not visible

#

so must have missing something in kernel

#

yeay it up 🙂

sharp girder
#

hello hi

#

I ran into an issue while booting from USB, but I found the solution and I am submitting a PR to fix it

#

is there a specific way I should do it (e.g. which branch should I merge into, PR format, etc.)

dusk phoenix
#

@sharp girder whoops missed that comment. Just create a PR for the dev branch

sharp girder
#

looking to make a PR which grabs the Apex driver source directly from Google instead of how it's obtaining it now, since the current version that's bundled with HassOS (1.0) is out of date (new version is 1.1) and seems to have significant performance improvements with Coral M.2 cards -- any suggestions on where to start? I can't find any reference to Apex at the moment either in operating-system or the buildroot repo, other than some build flags

#

here are the Coral docs for building it directly into the kernel as a module

dusk phoenix
plain steppe
dusk phoenix
#

@plain steppe without it docker load of the last image would fail. For some reason docker load requires temporarily more disk space.... In the end the effective image is the same size (reported by du, or even a couple of megabytes smaller)

plain steppe
#

gotcha. figured it was something dumb like that

dusk phoenix
#

Hehe yeah its kinda dumb. But it fits in that allocated 2GB image we have, so why not 🤷‍♂️

#

Also since we mount now with discard option set, it actually leads to a slightly smaller image afaik.

#

(as in, more sparse... In the end it probably doesn't matter since we write zeros anyways)

#

Actually was thinking about creating bmap images, so we could distribute sparse images one can flash faster. Probably not really useful for the average user, but can speed up flashing for testing etc.

plain steppe
#

oh cool i didn't know about bmap

#

i'm trying to sync our tree up to haos dev. we're still on 4.19 kernel and its causing problems with the new v1.4 rpi4 boards

#

lots of interesting changes recently 🙂

#

happy to see buildroot is a submodule now

#

also xz perf improvements. i was using pigz locally because gzip is so slow

#

lots of good clean up to hassos-expand

#

the new ha-cli looks very interesting.. where is the cli.sh its invoking onto tty1?

dusk phoenix
plain steppe
#

aha thanks, so ha banner does the info display

dusk phoenix
#

yes

sharp girder
dusk phoenix
plain steppe
#

any thoughts on periodic fstrim?

#

oh its already there in systemd

plain steppe
#

huh i'm noticing that after rauc install <file> the owner on the file changes to root

slim kiln
#

Hey! Is there a way to apply a patch to hassos manually?

#

(running on rbpi)

upper nebula
#

What kind of patch?

dusk phoenix
#

Ah wait, hardlinks share permissions. So should be fine I guess.

plain steppe
#

for rpi I2c, is the dtparam i2c1 or i2c_vc

#

doh different boards

#

am i remembering correctly there is something during rauc update that doesn't overwrite the old /mnt/boot/config.txt?

plain steppe
#

aha boot hacks in rauc-hook

upper nebula
#

Can someone explain me how release images are built? When I build image from dev branch than it shows developement images everywhere.
I see release branches not sure if that's the way it needs to be built

dusk phoenix
upper nebula
#

Thanks thats what I was looking for

plain steppe
dusk phoenix
#

Yeah that should work

#

There are other primitives in that driver it seems:

sd_printk(KERN_WARNING, sdkp, "sd_revalidate_disk: Memory "
              "allocation failure.\n");
#

but shouldn't matter

#

Are you sure the kernel got properly rebuilt and deployed?

plain steppe
#

it didn't the first dozen times but now atleast uname shows the correct version. I'll bump the version number too to make sure

#

thanks!

#

is there an easier way to do incremental kernel rebuild, vs rm -rf output/build/linux-custom

plain steppe
#

oh I didn't have the drive plugged in! I see the debug logs now yay

plain steppe
dusk phoenix
dusk phoenix
plain steppe
#

yea so we have been using consumer usb drives and they are becoming very slow after heavy usage

#

I found that many drives now are SMR, and it requires TRIM to allow the platter to garbage collect

#

but they don't use UASP so the kernel doesn't know it supports TRIM, or for readcapa16 drives it assumes WS16 for trim which fails

plain steppe
dusk phoenix
plain steppe
#

yea

#

it seems to assume all drives over a certain size will support writesame16 instead of checking the capabilities

plain steppe
#

here's example of two drives i have attached to my RPI. both are capable of TRIM, but one comes in as provisioning_mode=full and the other as provisioning_mode=writesame_16

#

if i change both to unmap i can run fstrim and it works

dusk phoenix
#

So how do you change them to unmap?

#

Ah I see udev

#

I mean we can start add those drives to udev rules I guess...

#

or is there someting in the kernel for device specific quirks or so?

rare sable
#

udev having such feature to doing that. It's also used for audio a lot. Most audio devices will not work correct on kernel without proper udev rules

wide meteor
dusk phoenix
#

Hi @wide meteor yes, its on my todo. I briefly looked at it already, I'll get back to you tomorrow.

#

Can I buy the device in the EU currently?

wide meteor
wide meteor
woven crypt
#

I running the HA OS in an VM on proxmox. I want to shrink the VM data disk. How can I shrink the disk inside the HA OS ?

plucky socket
#

@woven cryptyou would shut it down and do that from within proxmox

#

and this channel is for discussion of development of HA OS, not support

dusk phoenix
#

@wide meteor wrt cooperation I am the wrong person to discuss this. Please reach out to us using hello@home-assistant.io.

wide meteor
wide meteor
#

@dusk phoenix Hi. I have corrected the code according to your comments. As for the swap file, the question remains whether I create it in the right place?

eternal blaze
#

just a smol request: include netcat for people with haos on a pi with rescue trouble
to make sharing easier and reduce the amount of images landing in #330990055533576204
e.g.
mount | nc -N x0.at 9999
so people can share their output without pasting lots of images.

thank you for your attention

dusk phoenix
#

In the end both options are not that user friendly and ideal. Also they won't work in the HA CLI

dusky axle
#

although usually when users are on that stage they are already in the host console

#

and anything beats pictures of their glared up screens of the logs

dusk phoenix
#

The question is, will you really get the right thing if you ask them to type the above nc command? I mean, typos, wrong pipe... besides that they won't see the output of mount them self...

eternal blaze
#

in either case, a pin in the channel/psa in the forums/short description in motd about it would be nice for them to have, maybe a alias to which they can just pipe in their output...

#

alias sendtohelp="2>&1 | nc -N x0.at 9999" would be enough I imagine

dusk phoenix
eternal blaze
#

both work, just make it easy for them

dusk phoenix
#

Dumping /dev/vcs* might be interesting as well

hollow lintel
#

These types of sites change/stops working frequently.
We should not add that to the OS.
If we want to use something like that we need to run our own

eternal blaze
#

I can set it up for text only on that free oracle server atx mentioned if you want 😄 guess not

dusk phoenix
#

Not sure if its worth hosting something on our own 😅

hollow lintel
#

It is if we want this functionality on the Os

dusk phoenix
#

I fear the maintance burdon: abuse etc. for these kind of things are usually rather high

hollow lintel
#

I don’t disagree with that

dusk phoenix
#

I was considering using https://0x0.st/, but it seems down/DDOS'ed or something at the moment 😅

< HTTP/2 400 
< server: nginx
< date: Wed, 10 Nov 2021 15:39:50 GMT
< content-type: text/html; charset=utf-8
< content-length: 19
< 
Segmentation fault
hollow lintel
#

Exactly

eternal blaze
#

was just a suggestion, you decide about the need for this.

dusky axle
dusk phoenix
#

You just need to find the right width 🙂

#

I guess stdout redirect is better in most cases

dusky axle
#

but just did ha banner | curl -F 'f:1=<-' ix.io

dusk phoenix
#

Yeah, I tried to get a screendump. You can get the current ttys content from /dev/vcsX

#

Two aliases

alias send-to-ix.io="curl -F 'f:1=<-' ix.io"
alias send-screen-to-ix.io="cat /dev/vcs$(cat /sys/class/tty/tty0/active | cut -b 4-) | fold -w $(stty size | cut -d ' ' -f 2) | send-to-ix.io"
dusky axle
#

yeah that works

#

still gives me the weird output I opened an issue for 😅

fresh crown
#

I just tried ssh'ing into the host using port 22222 like I used to be able to but this time when I entered login it simply closed the connection. Did something change? I just wanted to look something up but I can't figure out how to get to the host filesystem anymore

latent wigeon
#

ssh will drop you at a shell prompt

#

Did you actually type ‘login’ while at a shell prompt?

fresh crown
#

I did, I didn't realize it was updated so that was no longer necessary. Cogneato got me sorted

#

Sorry about the cross-post, accidently released a version of an addon I maintain that was broken so was trying to fix it before other users updated

muted remnant
muted remnant
#

wow what a guy, he already did it

tight ember
#

I'm trying to enable host ssh to my HA Blue box (https://developers.home-assistant.io/docs/operating-system/debugging/#generating-ssh-keys) . I had this on the RPi I was previously using and it worked fine. However on the Blue it doesn't open port 22222 (port 22222: Connection refused). I tried deleting the authorized_keys file and rebooting to clear any existing keys. Then putting my id_rsa.pub file back on (as authorized_keys) and "import from USB" from Supervisor but the port is not opening. Supervisor logs show

[supervisor.os.manager] Synchronizing configuration from USB with Home Assistant Operating System.
21-11-27 12:07:37 INFO (MainThread) [supervisor.host.services] Restarting local service hassos-config.service

#

Does this procedure still work?

tight ember
#

I'm just getting started with the ability to do some development. I have a homeassistant_os install on an HA Blue device. I am trying to debug an exception that occurs in the aiohttp library on my production setup. I have installed the Remote Python Debugger integration and am able to connect to my HA machine.

I mapped Visual Studio Code to a git clone of core (and checked out tags/2021.11.5 to match my production code) and I can pause a thread and step through code. I need to set a breakpoint in a file in the python packages (/usr/local/lib/python3.9/site-packages/aiohttp/web_protocol.py). What's the best way to map that? Should I just scp the folder from my HA machine to my MacBook so it matches? Since the Python packages are dynamically populated via pip, I can't just clone a repo for that...

granite krakenBOT
tight ember
#

I ended up pulling over just the aiohttp subdirectory to my MacBook and mapping it locally. This is my launch.config 👆

#

I hit the breakpoint I needed! Now to debug...

plucky socket
#

@tight ember yes, it still works

#

i have debug access to my odroid/blue

spice zinc
#

Trying to build generic-x86-64 from the dev branch, but getting 404's when it goes to download intel-e1000e bits - anyone else seen similar?

#

Not sure if it's my build env, as this is an 'out-of-tree' module?

#

ah, url changed, will open PR 🙂

dusk phoenix
#

merged, thanks for the PR. The reason we did not see it on CI is because we cache the downloads 😅

unique sundial
#

Hi!
Does the the /boot/Config.txt file live somewhere?
I’m trying to rotate the display on my 7 inch pi touch screen.

#

I’m able to SSH into the host ok. Just not seeing a /boot folder…

unique sundial
#

Never mind. Found it.
It was in /mnt/boot

muted pawn
#

Greetings, hope the holidays have been good so far. Is there some sort of troubleshooting guide for moving the datadisk in OS 7? I've been trying to move the datadisk on a SeeedStudio Odyssey to an internal 500 GB SATA hard disk. The Odyssey has been working extremely well, and boots from a 120GB m.2 SATA SSD on an internal m.2 connector. I can see both the SSD (sda, sda1-sda8) and the HDD (sdb) when I go to the Hardware option in the Supervisor. But when I go to the 'move datadisk' option, I get a 'No suitable attached devices found' message. If I connect the Odyssey to a monitor and keyboard, I can partition the HDD, and have tried with no partitions, FAT, ext2 and ext4 partitions.... what else can I try?

#

I have also tested the HDD with windows, and I can partition, format and use the drive.

dusk phoenix
muted pawn
plain steppe
#

👋 working on updating our fork again

#

i think rpcbind is missing a dependency.. saw this on one boot:

Oct 12 08:24:07 dvr-server systemd[1]: Starting RPC Bind...
Oct 12 08:24:07 dvr-server rpcbind[198]: rpcbind: /var/run/rpcbind.lock: No such file or directory
Oct 12 08:24:07 dvr-server systemd[1]: rpcbind.service: Main process exited, code=exited, status=1/FAILURE
#

🤦‍♂️ never mind, BR2_PACKAGE_NFS_UTILS=y is something i added

plain steppe
plain steppe
#

i missed the HA Yellow announcements.. looks like a fun project

plain steppe
#

latest patchset which I confirmed fixes the fstrim issue on SMR drives not using UAS, like many WD Elements

plain steppe
dusk phoenix
#

Whops, missed those messages 🙂

dusk phoenix
#

fstrim on SMR drive: Interesting! Are those SSDs? Or is fstrim also important for non-SSDs nowadays?

#

Oh ok, read up on it on wikipedia now, magnetic drive which need fstrim. I don't think I had request for support for it on HAOS for it, but we'd like to improve support for separate media storage disks etc.

plain steppe
#

@dusk phoenix interesting.. if you can run our distro-diskinfo in one of those VMs and share output I can see if the virtual scsi is reporting itself as trim capable and if this patch would help

#

oh I guess sg3utils would be required also

rare sable
#

maybe something to os-agent that we can read it as property

spice zinc
#

Is GitHub actions for OS build using standard/shared runners? Notice build time is ~=30mins - trying to get similar with GitLab-CI…

dusk phoenix
#

No, its a hosted runner on a Ryzen 3900X machine

long shell
#

Hi there,

#

but after 2 hours no luck

#

I formatted an usb stuck to ext4, generated a key pair with ssh-keygen on linux

#

moved the id_rsa.pub to authorized_keys file on a stick calld "CONFIG"

#

It does not work. port 22222 is not even open now

ripe tartan
#

I'm also trying to get ssh access to the host system and failing to get it working. I generated the key with puttygen and made my usb drive in windows with NTFS formatting.

#

the "Import from USB" in the system settings UI doesn't seem to do anything either

#

Is there something I can look for in the logs or something?

#

Does it matter what kind of key I make (RSA, ECDSA, etc.)?

fast sluice
#

Hi,

Is there anyway to add a kernel module to the OS please ?
(I want to add ipset).

wicked kestrel
#

Hello!!
https://github.com/home-assistant/buildroot/tree/master/board/hardkernel/odroidxu4
that location has a boot.ini file, but the XU4 release has a boot.scr file instead of the boot.ini that is shown in the repo source. i was unable to find where this boot.scr came from (the contents look like RAUC but googling lines has not pinned it down) it has some extra jibberish characters (hopefully not a binary blob) in the head of the file.
The file is getting in the way of my boot environment as well. works with SD card but not my EMMC.

dusk snow
#

I'm looking for the kernel config file and any patches used to build kernel for odroid N2 and N2+, are these available? Tried a few of the links posted above they appear to be stale and don't include hardkernel N2 boards.

dusk snow
#

also any concept about what could cause systemctl status init-stage2 failed.

#

├─docker-99269a00000b14cc3718d354b435c180a7416ba30e5d1ba7564f2c6ce74b1643.scope
│ ├─9516 s6-svscan -t0 /var/run/s6/services
│ ├─9597 foreground if /etc/s6/init/init-stage2-redirfd foreground if if s6-echo -n -- [s6-init] making user provided >
!!!!!
init-stage2 failed.
!!!!!
│ ├─9598 s6-supervise s6-fdholderd
│ ├─9609 foreground s6-setsid -gq -- with-contenv backtick -D 0 -n S6_LOGGING printcontenv S6_LOGGING importas S6_LOGGING S6_LOGGING>
│ └─9750 sleep infinity

#

This is a straight debian bullseye system with supervised install. Kernel version 5.15.26. I'm trying to help the disto maintainer fix a few issues.

dusk snow
#

Just tried HA on the odroid N2 debian release and it gets the same init-stage2 failed error.

#

odroid debian bullseye release

violet rose
plucky socket
dusk phoenix
#

Working on it

violet rose
upper nebula
#

Any interest in adding support for radxa zero2?

shadow bolt
#

Is there any way to make sysctl settings persistent or would it be considered to at least increase the default rmem_max/rmem_default over Linux very restrictive defaults values of only 200K as I see there is at least one minor tweak for multicast. I maintain an addon that streams from cloud based cameras over WebRTC via UDP and, during initial startup, or when there are other small burst of CPU activity, the default buffer can be overrun in just a few hundred milliseconds and lead to significant video artifacts even though the resources are more than capable of handling the overall packet rate with slightly more buffering. Increasing the rmem_max/rmem_default, even to just 1MB, makes a huge difference in reliability and is a common tuning step for Linux when dealing with UDP traffic, but of course I can't do this from the addon container (at least, I can't figure out a way).

#

Sorry if this isn't the best place to ask this question, but it's not really a bug, just curious if things like this would ever be considered.

rare sable
#

Yeah, currently is no plan to have custom values for things like this (also hard to maintain with too much customizing on it). I guess we can talk about better value that we provide @dusk phoenix ?

dusk phoenix
#

Ideally we should pick a sensible value for our use case which "just works". Usually its worth checking what others are doing (generic distros, or specific distros like OpenWrt).

#

For the rmem_max/rmem_default case, at least my ArchLinux seems to use 200k as well. I am not against increasing it by default, if we can reasonably say that it has no major downsides ...

shadow bolt
#

Thank you very much for the for the response. It's fairly well known that 200K is very limited for modern, high bit rate RTP streams using UDP over fast links. Sure, for my Intel powered systems they can probably keep up just fine, but the poor RPi struggles quite a bit more. The default kernel settings are largely optimized for minimum memory usage with 100's/1000's of connections vs maximum performance of a small number of connections. There are lots of recommendations around to increase it to 25MB, but I think that is extreme. I've seen dramatic improvement in going from 200K to 1MB, minor improvement from 1MB to 2MB (during very high load with many streams), and little to no difference for values beyond that size. The only real risk I see is potential for slight increase in memory usage, however, socket buffers are allocated on demand from kernel memory and Linux has limits on total number of pages that can be used for all socket buffers (net.ipv4.tcp_mem & udp_mem) which are dynamically calculated during boot, so I think that risk is very low. However, I do understand risk of even minor changes for a project with so many users. I saw there already was at least one minor tweak for multicast so I hoped it might be considered. For now, I've managed to create a small udev rule file that users can drop into /etc/udev/rules.d as that seems to be one of the few places that allow for persistent changes. A high bar for some users, but at least an option. Thanks for your consideration.

fast sluice
#

Hi,
Is there any ETA tracking or similar on when PR merged on dev can be on master ?
I see that there is nftables and ipset in dev but not yet in master branch.

hollow creek
#

I want to upgrade to os 8.0 rc1 but get error: 22-04-08 06:40:19 WARNING (MainThread) [supervisor.jobs] 'OSManager.update' blocked from execution, system is not running - CoreState.FREEZE

random prism
#

Error connecting to GitHub

velvet sierra
#

Hey, I just noticed that the latest RC releases (8.0.rc1 and 8.0.rc2) in github are not marked as prerelease.
Is this a mistake or should I use somehting else to determine latest stable version (such as https://version.home-assistant.io/stable.json )?

dusk phoenix
#

stable.json is the ultimate arbitrator. I don't think GitHub has any influence other than displaying differently in GH releases

sage crown
#

Hi 🙂

#

I'm trying to build my own image

#

but my changes are not taken in account ...

granite krakenBOT
sage crown
#

do you know the differences between "image" and "images" ?

#

because only "images" links seems to be used to build the image

lavish thunder
dusk phoenix
#

Yes should still work

#

@lavish thunder note that you don't need to create a CONFIG directory if you use the USB flash drive import method

#

You need to name it (mkfs.ext4 -L CONFIG ...)

lavish thunder
shut pendant
#

Hiya folks -- I'm following the debugging guide at https://developers.home-assistant.io/docs/operating-system/debugging/ and have gotten my 22222 SSH setup just fine. But when logged into that SSH server, I don't seem to have access to docker nor systemctl etc. I know there was a change to have SSH open into a HA CLI container, but is there a way to SSH to the true host file system anymore? Like, I want to be able to inspect the actual core python packages.

hard spruce
# shut pendant Hiya folks -- I'm following the debugging guide at https://developers.home-assis...

I'm not really sure what the issue you are having is, unless you are somehow getting connected to the add-on ssh. The below is how it works for me:

C:\>ssh -p22222 root@homeassistant
Welcome to Home Assistant OS.

Use `ha` to access the Home Assistant CLI.
# docker --version
Docker version 20.10.9, build 20.10.9
# systemctl --version
systemd 249 (249)
+PAM -AUDIT -SELINUX +APPARMOR -IMA -SMACK -SECCOMP +GCRYPT +GNUTLS +OPENSSL -ACL +BLKID +CURL -ELFUTILS -FIDO2 -IDN2 -IDN +IPTC +KMOD +LIBCRYPTSETUP +LIBFDISK +PCRE2 -PWQUALITY -P11KIT -QRENCODE -BZIP2 -LZ4 -XZ +ZLIB -ZSTD -XKBCOMMON -UTMP -SYSVINIT default-hierarchy=hybrid
#
steep crane
#

I just had the second issue with the OS RC not restarting. On RC 2 I had to reinstall on RC3 I had to do remove power and reboot. Not seeing anything in the logs that seem helpful. Is anyone else seeing this on the RPI4?

dusk phoenix
#

Not so far. So it got stuck during reboot? Anything on the screen? Is that using USB SSD?

steep crane
#

It is usb boot. I posted the logs last time. This time I was away from the house when I updated and the browser did not load. Nothing showing on the PI screen. I removed power and everything came back up. I asked this question as the last 2 RCs did not come up correctly for me after the update.

plucky socket
#

I've had no issues with rc2 on a pi4 but it is not using external boot

thorny viper
#

Is anyone working on support for the ODROID-M1 yet? If not, what would be the process to get cracking on it? I've read https://developers.home-assistant.io/docs/operating-system and just wondering where to start, since we already have other ODROID images. I tried all the other ODROID images (N2,N2+,C2,C4 and XU4) and none of them worked (petitboot threw exception and exited to shell during device scanning) on the M1.

#

I've never worked with machines that require DTBO files, so some docs on how all that works would be needed too.

dusk phoenix
#

I think I saw a first patch adding support for the board in upstream Linux on the ML recently, but it was lacking PCIe afaicr

dusk phoenix
#

Probably the biggest challenge is U-Boot. We use upstream 2022.01 U-Boot for all supported boards, and ideally like to do so for every new board as well. Basically, the steps are: Creating a new board directory in hardkernel subdirectory, figuring out if and how upstream U-Boot supports that board. If yes, potentially patch U-Boot and build it for the target, if no, try using a dowstream U-Boot (not really preferred!). Next up the kernel, it seems that the SoC is supported, but a custom device tree needs to be used which is not part of 5.15 we are using. Hence this needs patching of the kernel. Then make sure the dtb gets build and deployed. This should be fairly straight forward.

thorny viper
#

@dusk phoenix Thanks a bunch for the feedback. Now I at least know how to approach things. Not sure if I'll find time to work on it, but I know where to look.

versed axle
hard spruce
steep crane
#

I just tried RC 4 and it doesn't look like it is coming up. I'm at 15 minutes and nothing is showing on my attached monitor. I'll wait another 1/2 hours and see if anything happens.

dusk phoenix
#

That is with SSD right? It must be related to the firmware or U-Boot not being able to enumerating the USB SSD properly

#

What type of SSD controller and SSD is that? Does reboot work normally?

#

What kind of power supply are you using?

steep crane
#

This is an SSD Kingston, with Startec 3.1 controller. The reboot does not work, screen is blank, screen works without SSD plugged in. Power supply is rated at 4A and I have the SSD on a hub. I have had issues which each of the RCs. I do not believe this is a power issue.

dusk phoenix
#

But reboot did work with 7.6?

steep crane
#

It did

#

I just reinstalled via etcher and it is booting now

#

I will try the update via cli when I finish the restore. This is what worked last time. It did not work via the UI

dusk phoenix
#

So powering it off and back on does not help?

steep crane
#

It left it with a blank screen. On one of the RCs ( I believe 3) it did. On one, I had to do a restore.

dusk phoenix
#

Ok, but before doing a restore, in the current state, when you remove power and put it back, does it boot now? Or really nothing, just blank screen?

steep crane
#

I removed power twice and both times the screen was just blank. I am doing a restore now.

#

I checked to make sure it wasn't a screen issue and tried it without the SSD attached and the screen came on.

#

Also there is a restore issue. I do not see the onboarding option.

dusk phoenix
#

I just updated a bunch of Raspberry Pis (specifically, a 2xRPi 4 with SD card, 1xRPi 3 Plus with Samsung USB flash drive, and RPi 3 with regular SD), so each one of them came back online no problem, so it is definitly not a generic issue.

#

It is some Firmware/U-Boot <-> USB SSD interaction (once again)....

steep crane
#

What would you like me to try?

#

I would hate to have you release 8 and see the issues that effected a small group of us for almost a year.

dusk phoenix
#

Are you familiar with USB to UART (TTL) cables for Raspberry Pis? Do you have one?

dusk phoenix
#

I mean, specifically for OS 8.x

steep crane
#

I had an issue, but closed it as I thought it was on my end (operator error). I have had issues with each one. After the the last restore I installed from CLI and it worked. After the RC3 I powered off and back on and it worked.

#

I do have a usb to uart that I use to flash devices. Not sure if this is what you are looking for?

#

HiLetgo CP2102 USB 2.0 to TTL Module Serial Converter Adapter Module USB to TTL Downloader With Jumper Wires
HiLetgo

dusk phoenix
#

Yeah probably works. So you can reproduce the issue right? When you restore with 7.6 now, and update again, it will happen again?

steep crane
#

No idea. I am still restoring. What I was going to try next was CLI as this worked last time. I can do what ever you think is best.

dusk phoenix
#

Yeah I think a good approach is to make sure you have a workflow which reproduces the problem. Ideally it should be a "minimalistic" work flow (with as few as possible external variables). From how you describe it, (hanging at boot), it should not really be related to your installation. So that begs the question: Is it also happening without restoring first? Just start a new instance and update right a way?

steep crane
#

I have already restored everything. I'll try this next time.

#

should I do cli or UI?

dusk phoenix
#

Shouldn't really matter, probably CLI is better since you would get direct feedback in case the update command fails

#

I test now with a StarTech.com USB3S2SAT3CB & Intel 520 Series S-ATA Disk.

steep crane
#

Do you want me to try directly to RC 3?

dusk phoenix
#

You mean installing RC3 or updating?

#

Ideally just update to rc4.

steep crane
#

OK. Trying from CLI now.

#

I don't think you see any more here as from the cli, it says system shutting down.

#

I think it is working from CLI as the screen says waiting for supervisor

#

maybe I talked to soon as the screen is now blank

#

No activity on the ssd or screen

#

What now?

dusk phoenix
#

You can connect that SSD to your PC correct? Can you find the config.txt on the first partition and make the following adjustments:

steep crane
#

I have the file open

dusk phoenix
# 
enable_uart=1
#

(uncomment it)

#

Which operating system are you using on your PC?

steep crane
#

windows 10

dusk phoenix
#

The speed setting must be 115200

steep crane
#

Not understanding. You want me to connect the pi to the adapter and the adapter to the PC?

dusk phoenix
#

Yes

#

So: PC <= USB => Adapter <= UART TTL => Pi

steep crane
#

So no USB to SSD?

dusk phoenix
#

Once config.txt is adjusted as above, save the file and eject the disk from your PC and connect it back to the Pi

dusk phoenix
steep crane
#

With the config.txt change the pi looks like it is booting

dusk phoenix
#

Hm

steep crane
#

It is booting and my screen is loading

dusk phoenix
#

Did you have any other config.txt options adjusted or added?

steep crane
#

I only changed the one item you said to change. Whether there was others I have not idea.

#

Everything is up again.

dusk phoenix
#

Can you reboot 2-3 times to see if that really is reproducible?

steep crane
#

Do you mean restart or reboot the pi?

dusk phoenix
#

What's the difference? 😅 Software reboot

steep crane
#

One turns off the power, the restart only restarts HA?

#

I'll power down and restart.

#

any idea where they moved the shutdown on the new menu?

#

doing it from CLI as I can't find it in the new menus

dusk phoenix
#

System -> Hardware

steep crane
#

It did not come up first time. I removed power twice and then it rebooted.

dusk phoenix
#

That is so weird

#

That hardly can be a software issue...

steep crane
#

I'll try it again

#

What did the enable uart do?

dusk phoenix
#

It enables log output on the serial console (UART). The Raspberry Pi firmware tells the bootloader (U-Boot) and the subsequent Kernel to show a console on TXD 14/RXD 15 GPIO pins

#

But it shouldn't really matter

steep crane
#

So something is staying charged (stored) that is clearing after power removed for a short time. I am going to unplug the pi for 10 seconds and then power on to confirm my theory.

#

It boot right up

#

any chance a register is not being flushed?

#

reboot 2 more time (with power unplugged for 10 seconds) and it worked each time.

dusk phoenix
#

It is unclear why I cannot reproduce this

#

It is a Raspberry Pi 4 correct? How old is the device? Maybe it depends on the firmware flashed on the RPi4 itself.

steep crane
#

RPI 4, 4 G. I was one of the people that had upgrade issues for almost a year (1119). I have not reflashed the device in most like a year. I'll be out for a while. If you need more info, let me know and I will get it to you later today.

dusk phoenix
#

I do remember and recognize you, but I didn't remember anymore if it was indeed RPi 4 🙂

sly beacon
#

Hi,

I am not sure if this is the correct place, but I couldn't find better topic for this.

Since I upgraded to 2022.5, (QEMU image running on Proxmox VE), HASS is randomly restarted by the Supervisor.

Here's the log messages of the Supervisor:
**
22-05-09 21:33:42 ERROR (MainThread) [supervisor.homeassistant.api] Error on call http://172.30.32.1:8123/api/config: 22-05-09 21:33:42 WARNING (MainThread) [supervisor.misc.tasks] Watchdog miss API response from Home Assistant 22-05-09 21:36:44 ERROR (MainThread) [supervisor.homeassistant.api] Error on call http://172.30.32.1:8123/api/config: 22-05-09 21:36:44 ERROR (MainThread) [supervisor.misc.tasks] Watchdog found a problem with Home Assistant API! 22-05-09 21:36:44 INFO (SyncWorker_5) [supervisor.docker.interface] Restarting ghcr.io/home-assistant/qemux86-64-homeassistant 22-05-09 21:40:24 WARNING (MainThread) [supervisor.homeassistant.websocket] Connection is closed

Tried to search online, so many similar symptoms, but different cause.

172.30.32.1 is the IP I see on eth0 when I ssh to HASS, but its not the IP address of HASS and its not even the subnet of my internal network - so not sure what is the purpose of this address...

dusk phoenix
lapis tiger
#

After I have fallen into the Case sensitive State Pit from making an Automation with UI - is it somehow possible either to ensure correct cases or simply make all states Uppercase?

steep crane
#

Not seeing the 8.0 release. By the way the # homeassistant_os says archive channel . Is this still used?

oblique remnant
#

hey, I've been reading the rauc docs for an hour and a half and I can't figure out how I can use it to upgrade my hassos device to a dev build I just made. Isn't there some way to just disable the signature verification when I'm doing an explicit rauc install from the OS shell?

#

and I've copied the certificate over and tried using --keychain=../path/to/cert or --conf=../path/to/config (with the config setting the keychain) but I still always get this error:

# 
 40% Verifying signature failed.
 40% Checking bundle failed.
100% Installing failed.
LastError: signature verification failed: Verify error:self signed certificate
dusk phoenix
oblique remnant
#

no worries

dusk phoenix
#

We do have a dev CA, I should be able to create a cert for you

oblique remnant
#

I figured out a workflow that's okay: I can copy the squashfs over (via ssh or similar) and use rauc write-slot (or whatever it was called) to put the rootfs (and kernel, if changed) into the other slot

#

and then switch over with rauc status

dusk phoenix
#

Hehe, yeah I guess that works too 🙂

oblique remnant
#

that was good enough for what I was doing anyway 😉

dusk phoenix
#

Ok, yeah for rootfs replacement its probably quite straight forward. I guess you also have to replace the kernel modules match

#

Feel free to PM me in case you want a cert

oblique remnant
#

you mean when the kernel version changes?

dusk phoenix
#

Yeah or signature... e.g. when adding drivers.

oblique remnant
#

I did a manual upgrade (backup, reflash all, restore) to dev on my first attempt

#

after that I flashed only the rootfs... I think?

#

and i added a bluetooth module

#

might remember wrong though

oblique remnant
#

Wifi is working perfectly for me now with that PR + the upstream fix for usb_modeswitch (haven't sent it to the buildroot mailing list yet)

#

generally, do you patch/add things in buildroot-external while getting them merged upstream?

#

I didn't really think about it yesterday, but the rtl8821cu driver could be upstreamed as well (if they'll take it)

#

and the usb_modeswitch fix could go in buildroot-external in the meantime to have the dongle work out of the box for the next haos release

dusk phoenix
#

I do have some patches on buildroot/ itself, those I try to upstream as much as possible.

dusk phoenix
#

Is there no firmware needed?

#

Oh I see, you are talking about `usb_modeswitch which is upstream. We can add it to our fork for now, but please send the patch upstream.

oblique remnant
#

okay, should I put it together with the bluetooth module?

dusk phoenix
#

You mean in one PR? Fine by me, just separate commits

#

Or do you mean the rtl8821cu package? I would leave that in buildroot-external, since its a new package

oblique remnant
#

ah I see, okay

#

then that's two PRs, sure

oblique remnant
#

the other thing I've been working on is bluetooth audio in haos, so far I've seen only posts in the forums etc. doing it on supervised installs

#

actually the only real "low-level" change it needed for me was enabling BR2_PACKAGE_BLUEZ5_UTILS_PLUGINS_AUDIO

#

at that point I could pair and connect to a bluetooth speaker and pulseaudio would see it

dusk phoenix
#

Cool, yeah haven't looked at Bluetooth audio so far

oblique remnant
#

I had to run ha audio reload to make the devices show up in the UI though

dusk phoenix
#

Nice! Would Supervisor pick it up?

#

Oh I see, yeah ideally we should automate that too. I guess you used bluetoothctl directly?

oblique remnant
#

the problem might be to delay addon starts if they are configured to use that device

oblique remnant
#

or maybe pulseaudio can somehow make a dummy device until the real output shows up?

#

that would be preferrable I guess. definitely a lot to figure out still to make it a smooth experience

dusk phoenix
#

Yeah not sure how pluggable devices work exactly with Pulseaudio. I'd guess udev is involved, since its all Lennart stuff 🙂

oblique remnant
#

udev for alsa as far as I know, but bluetooth is separate

#

theres two pa modules (module-bluetooth-discover and moodule-bluez5-discover)

#

and they just dynamically add/remove the sink... but thats a bit of a problem if someone relies on being connected to a specific sink on start

#

there may be pa modules to deal with this sort of situation, don't know

#

right now basically when I start the system, I have to open bluetoothctl and initiate the connection, then run ha audio reload and restart any addons that are configured with bluetooth output

#

at that point it works great

dusk phoenix
oblique remnant
#

after something like 2h bluetooth disconnects but I had a look at the logs last time and it actually looks like the cheap USB speaker I also have somehow crashes and takes down all of pulseaudio with it ^^

dusk phoenix
#

There is a NewCard signal

oblique remnant
#

oh yeah its definitely possible to notice and add/remove a dummy sink

#

I rather wonder if/how its possible to have the other application not notice/care

#

but I dont know much about how the addon audio integration is handled right now

#

I've seen the supervisor creates some config files and mounts them into the containers

dusk phoenix
#

Yeah also not that familiar with that code base tbh.

dusk phoenix
oblique remnant
dusk phoenix
#

General plug ins

fading nimbus
#

Hi, I just want to thank the Home Assistant developers for providing Home Assistant Operating System 8.0, including the linux kernel 5.15 that has the ntfs3 driver with write support for NTFS partitions. Here is my short and grateful story:

Today I tried using some udev rules to automount my pendrives into the media directory (using https://gist.github.com/eklex/c5fac345de5be9d9bc420510617c86b5) and I found out I could not write to the drive. I realized it was because the ntfs kernel driver provides read-only support, and I remembered that there was a new ntfs3 driver available on Linux 5.15 (https://www.phoronix.com/scan.php?page=news_item&px=NTFS3-For-Linux-5.15). I upgraded to Home Assistant OS 8.0 and, with minor changes to the udev rules, I could adapt them to select the ntfs3 driver if appropriate instead of the ntfs driver (https://gist.github.com/zeehio/7c67d531236e39496c39b3c5c3b3f268).

I was very fortunate to try this just today, a few days after the release of Home Assistant OS 8.0 on github. I guess you may not consider this ntfs3 feature to be relevant for most users, but it has been very convenient for me.

Thank you all for your work.

rare sable
dusk phoenix
rare sable
#

yes

#

but also that have an event system

#

and we have an clickup to feed this events into our event bus like docker events

#

from pull to push

dusk phoenix
rare sable
#

yeah, the library is new, we have also an clickup to replace the library

dusk phoenix
#

@oblique remnant so it would mean getting into Supervisor development 🙂 💪

oblique remnant
#

What is a clickup?

dusk phoenix
#

That is a task management system we use at Nabu Casa.

oblique remnant
#

what you're saying is, the supervisor should trigger audio reload (internally, not via CLI) when the pulseaudio devices change, right?

#

btw, i've noticed that once an addon is started with a bluetooth speaker, I can turn that speaker off and on and it reconnects smoothly 🙂

#

i haven't actually tested if it works even when the addon is started without the speaker connected, if we're lucky it might even

dusk phoenix
dusk phoenix
#

And pulseaudio under the hood picking that up as well.

oblique remnant
#

yeah, there's probably some pulseaudio module loaded that does the fallback behaviour

#

moving streams to a dummy sink/source and back

blissful quest
#

@latent wigeon i use 'login' and looked around and found /mnt/data/supervisor/tmp/ had 5 directories. 18.5G 2.8G 213G 8K 41G totalling 275G in the Host OS. Two from 2021. With what seems parts of backup. <adonn>/tar.gz or tgz i forget. cleaned out and now i have space back. Thanks.

lucid stone
#

hello.. I have been seeing increase in cpu util from the hass proc running in venv, is there a way to debug this to see what HA is busy with.. this is a rather busy installation with large number of integrations / devices / automations.. etc.. e.g. now if I go to developer tools, it takes upwards of 10-15 seconds to load the page..

rare sable
lucid stone
#

as a best practice u mean ?

rare sable
#

that is what we support, running somethings different is not a good idea and you will also not get support for this

#

btw, this is the developer channel not support channel

lucid stone
#

oh okay I follow.. its been running well for 4 years so I didnt convert to the OS .. I ll give it a go .. thank you ..

tardy gale
#

On TrueNAS 13.0 - does anyone know how I update to Cargo edition 2021?
Before I update I'm not able to update HA to 2022.7.*

subtle escarp
#

Is there any os testing channel

long thorn
#

Sorry if I am at the wrong subgroup, but my config directory shows amongst others the following entries:

-rwx------ 1 jan jan 0 jul 19 10:32 home-assistant.log.fault
-rwx------ 1 jan jan 466665472 jul 19 12:34 home-assistant_v2.db
?????????? ? ? ? ? ? home-assistant_v2.db.corrupt.2021-10-17T02:12:00.805531+00:00
?????????? ? ? ? ? ? home-assistant_v2.db.corrupt.2022-01-18T16:20:15.695854+00:00
-rwx------ 1 jan jan 65536 jul 19 12:34 home-assistant_v2.db-shm

Apparently a problem with the file system (a SSD drive hooked up to a PI4).
I can I repair this ?
Thanks in advance, Yann

dusk phoenix
dusk phoenix
abstract bison
#

how do i configure z2m in the new os8.3?

#

or where may i ask?

#

i always get "no need to configure"...

latent wigeon