-
Notifications
You must be signed in to change notification settings - Fork 149
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Issues with building Armbian images #7
Comments
Fix extraction of Armbian images (part of #7)
Strange, it worked for me with the ExampleOS. See #2 where @huelvayork claims it works for them too. |
Well, I had to at least change the The fix implemented in #8 doesn't harm any of the implemented at least, so it's good it's merged. Maybe you could try to build an image for one of those Armbian images? You will probably see the issues I experienced as well. |
I can't really test those images without the devices. |
Ok, I wasn't aware of the Meanwhile, you can test any built image from your computer as well. I've booted into mine from qemu without any issues, and reproduced the login issue. I'm not saying you should, but this is how it could be done. |
Next issue:
I am not quite sure yet why this occurs. This is the Ubuntu build. I'm going to fix it soon. If you didn't guess already, I should mention I am trying to build OctoPi. |
How are you booting from qemu? |
It's not very stable, and it's quite some work (you need to extract and provide the right kernels to qemu etc.), I'll have to test that a bit more in order to be able to provide instructions. In the meanwhile, I found out what issues I had with the As soon as I get to figuring out how to use the damn |
Hey, I manage to run builds now using both Ubuntu and Debian Armbian builds. So if no problems I am going to close this. |
fine!
2017-12-26 9:20 GMT+01:00 Guy Sheffer <notifications@github.com>:
… Hey, I manage to run builds now using both Ubuntu and Debian Armbian
builds. So if no problems I am going to close this.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#7 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAqjPxzsqhwCfW_tyS8oQrn239DjAt-Nks5tEKxTgaJpZM4P5JbF>
.
--
Un saludo,
David Prieto.
|
Was waiting for @TheAssassin to reply who is OP. Closing if no reply |
Ah. I didn't think I would have to take any further action. I'll test it and report back. |
Well the issue is open, but it's also a free open source project. So I don't know if it's works unless test it. We are volunteers here :) |
There's information missing how to build Armbian images (i.e., what to call with which parameters). I think https://github.com/guysoft/CustomPiOS/wiki/Building needs some information on that. What I did was:
Seems to work now, the image is extracted, I was quite happy about it. However, right after the
Re-calling this script with At least I got a file called As you might have seen, there's a few minor issues which are still left. But it's much better than when I first tried it, so thanks for improving this so far. |
The orange Pi Zero image for example is called `Ubuntu_xenial_next.7z`, hence the pattern here doesn't match. Changing the glob mask fixes this issue. CC guysoft#7
We could add a page in the wiki with this info. Because ot seems there is a guide needed for armbian builds |
Yeah. The "actual" official build guide is in the wiki, too (although that should IMO be moved into the README, or be highlighted better in the README, I had some trouble finding it). |
We can link to it from the readme |
... or move this to the README. I guess this is what everyone is interested in, it's literally the most essential part of the documentation of this project. Half of this is in there anyway, so moving it to the README entirely would eliminate those duplicate docs, too. Also, in my experience, it's easier to keep a README up to date with the code than a wiki page, everything's available offline from the same directory. |
Ok, will also accept a pull request for the readme. |
Well, you can still link to the README from those places, it doesn't really matter whether it's in the wiki or in the README. But well, I'll stop argumenting about semantics. I'll try to send a PR. |
Hi, I think I will share my exeprience here. I'm trying to build a distribution based on Debian based armbian for an Odroid C1. After installation, logged as root:
Then logged as octopi:
python 3 is not available by default, I had to add it to avoid build error, with: Then the strange thing is I cannot launch the build without sudo, or I get errors of fsdisk not available. img is fully created, the odroid c1 boot, but I can't get it by network (no DHCP lease requested). I could go to the Ubuntu flavor, but I prefer if possible to stay on headless Debian. |
Regarding the adduser issue, I think it's implemented by this script: So probably it could be bypassed by removing the file /root/.not_logged_in_yet, does that make sense ? |
@ChrisP-Git I remember a similar issue, which was solved by installing I won't get to testing this with real hardware any soon, but I will let you know once I do. Maybe check the commit history? All that is needed is something to install the |
@TheAssassin thanks for the head-up. I tried to find a PR concerning adduser but didn't find any (maybe I missed it). The issue I have now is that eth0 is not started by default. tried something on a fork but it didn't fixed it. |
The idea is to get rid of the default username/password combination. There's 1000s of similar devices with SSH servers available on the internet (via port forwardings). The only way to get rid of those, though, is to force people to add another user with different values on first use. A lot of users would otherwise not change either value for convenience. It is explicitly recommended to not "solve" this problem like you suggest, but rather fix the dependency issue. |
I get your point, and preventing end user to simply redirect ssh port is a way to go. |
I suggest we start with a basic guide how to do it at all and then expand it. Before changing the default security in Armbian. |
Well, it's a bit tricky as the standard initialisation phase on Armbian should be to logon as root, change the root password as requested, then create a new standard user. However for OctoPi a prepopulated user (pi) is required to run the OctoPrint processes then created by the build process. And this is the process for the standard Raspbian image, and the others variants, right ? So should this creation of user 'pi' inside the build be reverted and finding a way to use the user created during first run to execute the OctoPrint processes ? I'm not saying that a way is better than the other, but I am a bit confused. |
@guysoft to users of this project, it won't matter whether the default Armbian behavior is changed or not, in my opinion. I would say they prefer it to be consistent across distros, actually. |
I've read this thread and the previous one, and i'm very confused what the status of this is. |
Just tried to build for Armbian.
and for some reason, pip fails:
|
@indrora Can you please provide the base images you are using to build? |
It's Debian for Orange Pi Zero:
https://dl.armbian.com/orangepizero/Debian_stretch_next.7z
The DNS server issue isn't tied to a specific image, so I'll go put in a PR
for that.
On Mon, Jan 28, 2019, 01:56 Guy Sheffer ***@***.*** wrote:
@indrora <https://github.com/indrora> Can you please provide the base
images you are using to build?
On Jan 28, 2019 01:56, "Guy Sheffer" <notifications@github.com> wrote:
@indrora <https://github.com/indrora> Can you please provide the base
images you are using to build?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#7 (comment)>,
or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AAcuOLPajp8U6c7irZFA4ITXaVwkLXQnks5vHrtEgaJpZM4P5JbF>
.
|
Reference on Pip: This is a known thing and we should not be installing Pip like this? see here: pypa/pip#5599 |
Just noticed something: Pip gets installed, upgraded, virtualenv installed, then pip immediately removed. This leaves things in an inconsistent state. This generally isn't safe and it's generally safer to use the distribution provided version of pip, virtualenv. |
@indrora I can change it to:
Would that work? |
Looking at it, I think the pip install was there from either legacy code or something with the iptables, I suggest you just remove all of that, confirm it works without it, and then we commit that bit out. |
@guysoft I have to remove NetworkManager to get wifi work becuase NetworkManager control network after armbian buster. ( armbian has option for changing boot filesystem other than ext4 |
@luxflow So since this issue has been open. I've been playing around with the idea of making armbian stuff as a variable rather than a variant (and have the variant just set the variable). In latest devel there is support for Ubuntu like this.
There is a new variable
Then we could have
and
Do you have an image you are building? I could test it and start implementing the new way to build, which should be easier to maintain. |
@guysoft https://github.com/luxflow/CustomPiOS/releases/tag/buster |
@luxflow Cool, we should really merge your changes in to mainline. |
@guysoft So why not disabling it? I will PR soon |
Fedora Server uses NetworkManager on ARM as well, as does CentOS. They've
advised against networkd for a lack of maturity.
Raspbian uses others because of any number of reasons, including long
standing compatibility things?
I would advise against removing it, given other distributions usage.
…On Sun, Jan 26, 2020, 21:00 luxflow ***@***.***> wrote:
@guysoft <https://github.com/guysoft>
IMO, network manager is useless for our case
As I know, it is originally intended for laptop or desktop. raspbian don't
use NM by default
armbian use NM because it is good choice for inexperienced users.
https://forum.armbian.com/topic/6104-network-manager-woes/
armbian has NM based headless "onetime" wifi setup (armbian_first_run.txt)
***@***.***#diff-d41311dfd21fdaa3fcceec5294fa29acR34
<luxflow@2979c8a#diff-d41311dfd21fdaa3fcceec5294fa29acR34>
but I prefer wpa_supplicant.conf method which can change wifi setup
multiple times
So why not disabling it?
I will PR soon
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#7?email_source=notifications&email_token=AADS4OHGCI3YQT73STZ35RLQ7ZS6PA5CNFSM4D7ES3C2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEJ6J2ZY#issuecomment-578592103>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AADS4OHZTRS5FX44MU3P5B3Q7ZS6PANCNFSM4D7ES3CQ>
.
|
systemd-networkd seems off-topic thing isn't it? and I found why armbian drop fat boot partition and use ext4 by default personally, I think we should have to deal with all of these problems from a practical point of view
I'm not sure which way is better |
I rather stuff stay similar to the Raspberry Pi initial configuration (well with the CustomPiOS edition that is part of the network module, such as having a wifi text file with presets).. So I vote:
|
#68 just went it, it might solve some of your issues. |
I tried to build an Armbian image. As suggested by the
variants
directory, this is officially supported.Issues I noticed:
adduser
command is not available, preventing a console login as a whole. I have not invested too much time into debugging this, and am now running an Ubuntu build, which I hope will work better. But at leastadduser
should be installed by your scripts..7z
). Therefore, I replaced theunzip
call with a7za
call. This should work fine for.zip
archives as well. I'll create a PR soon.Regarding the missing dependencies, someone has to invest some time into testing this, which I don't have at the moment. PRs welcome I guess.
(If the Ubuntu build shouldn't work either, I'll considering taking the time for further tests with Debian as well, but will try to fix at least the Ubuntu build.)
The text was updated successfully, but these errors were encountered: