Lots of things seem to be broken.
g_multi is flaky, doesn't connect via usb0 to the host.
Then when connecting to the BBB via its network address over ethernet:
Cannot GET /%7B%7Bsite.baseurl%7D%7D/Support/bone101/
Then connecting to the ethernet IP:8080, gives and empty apache directory.
When first setting up the latest testing image, it took me a bit to figure
out how everything needed to be setup, and finally I had to edit the boot
partitions uEnv.txt and manually set ${uname_r} variables to hard paths. Not
sure if this is normal or not, but I knew this would work. Also the first
This isn't normal, you shouldn't have to touch anything.
several times I tried to log in via ssh with puTTY or Linux ssh
root/debian@ipaddress, I could not log in.
odd, it does take up to 30 seconds on the "first" boot for the ssh
key's to regenerate.
Overall I would have to ay the experience was rather disappointing. I had
hopes of toying around with the cloud9 version that runs on Nodejs v0.10.x,
but no joy . . .
Any chance of us normal people getting the source for the version of cloud9
that does run on Nodejs v0.10.25 ?
Complain to the cloud9 people... This was sup post to be "open
source'd" around april
Regards,
Lots of things seem to be broken.
g_multi is flaky, doesn't connect via usb0 to the host.
Can you give some details?
Then when connecting to the BBB via its network address over ethernet:
*Cannot GET /%7B%7Bsite.baseurl%7D%7D/Support/bone101/ *
I didn't realize Robert was pulling from the head of the tree. It needs to
be pre-processed with 'jekyll build' and then pull the contents from the
_site directory. I've been looking to see if I can get the webserver to do
the Jekyll pre-processing itself. Jekyll was used because of its
integration with github-pages.
Then connecting to the ethernet IP:8080, gives and empty apache directory.
When first setting up the latest testing image, it took me a bit to figure
out how everything needed to be setup, and finally I had to edit the boot
partitions uEnv.txt and manually set ${uname_r} variables to hard paths.
Not sure if this is normal or not, but I knew this would work. Also the
first several times I tried to log in via ssh with puTTY or Linux ssh
root/debian@ipaddress, I could not log in.
You mean, you needed to wait for the device to boot and the server to start?
Overall I would have to ay the experience was rather disappointing. I had
hopes of toying around with the cloud9 version that runs on Nodejs v0.10.x,
but no joy . . .
Browsing to http://<IP>:3000 didn't work for you?
Any chance of us normal people getting the source for the version of
cloud9 that does run on Nodejs v0.10.25 ?
You can always extract what is in the /opt/cloud9 folder.
http://github.com/c9/c9v3 is still private, despite them telling me they
had plans to open it up many months ago. You can ping the owners of that
repo.
You mean, you needed to wait for the device to boot and the server to start?
No, I mean I got “invalid login” when trying to login via ssh. Period. Perhaps it could have been the key generation Robert mentioned above. As I had to make an adjustment to the interfaces file( from the host ). When I rebooted, it worked.
Browsing to http://:3000 didn’t work for you?
NO port 80, and 8080 only. I did not try port 3000 as the error i pasted above indicated that Nodejs was somehow listening on port 80.
Also i can not elaborate on what was happening with g_multi. It just wasnt working. A Linux host was not loading cdc_ether, and a win7 host was not prompting for drivers. The module was loaded according to lsmod though, and removing the module was impossible.
Afterwards, I replaced the rootfs with Robert barefs off his build instructions, and made a lot of “radical” tweaks to the file locations, ripping /lib/modules out from the stand alone image and adding to the barefs then uEnv,txt to make it all work. g_serial works fine, as i had to use it to pull in openssh-server.
Anyway I’ve been up all night messing with this will test more later
Port 80: is bonescript (nodejs)
Port 3000: is cloud9 (nodejs)
Port 8080: is apache2
Regards,
You mean, you needed to wait for the device to boot and the server to
start?
No, I mean I got “invalid login” when trying to login via ssh. Period.
Perhaps it could have been the key generation Robert mentioned above. As I
had to make an adjustment to the interfaces file( from the host ). When I
rebooted, it worked.
Browsing to http://:3000 didn’t work for you?
NO port 80, and 8080 only. I did not try port 3000 as the error i pasted
above indicated that Nodejs was somehow listening on port 80.
Port 80: is bonescript (nodejs)
Port 3000: is cloud9 (nodejs)
I think it might be worth me seeing if bonescript could be more tightly integrated with cloud9 as an option such that cloud9 could move to port 80. not sure what those guys use for HTML content preprocessing.
Jason, so now that I’m rested a bit and less frazzled, I can elaborate a little more on the g_multi issue I was having.
I check dmesg, it loaded apparently according to the log. All 3 devices were listed. The g_ether portion of the module seemed to bounce between eth1, and eth2. issuing ifconfig ethx up would bring the interface up, but it was not taking an IPV4 address. Is there a connection manager in the latest testing LXDE image thwarting my progress ? I’m not too familiar with X tools s such as I tend to stick with CLI only.
The inability to unload g_multi stems from the mass storage portion mounting a file system apparently and it not related to this specific kernel / image as far as I know.
It’ll only take me around a half hour to dd the partitions back over later one today, and I do plan on doing that. I just wanted to toy around with “hacking” a boot / rootfs together to see what I could get away with.
I did check port 3333, as that is what I had set for my own cloud9 build that never realty worked. But the port I used was based off the instructiond on the cloud9 git as far as i remember.
I’ll definitely look into hacking in that cloud9 build, and it seems I’ll be building from scratch somewhat anyway. As my hackfest yielded less than optimal results. The barefs the way I set it up was 108M, and with openssh-server around 144M
Jason, Ok cloud9 seems to be functional, and I have to say this is pretty awesome. Where can I find documentation on how you have this setup ?
Ok moving on to the other things for this image. First, there is no uEnv.txt file period. We have bbb-uEnv.txt, and nfs-uEnv.txt. Neither one of these will work, because I have the original uboot from an A5A. So, I follow Roberts build from source instructions, so I know how to deal with this. Someone else may not, but this is a testing image after all - I guess.
I’m still poking around, but I disabled HDMI in uEnv.txt, and the image now seems to be more responsive. networking interfaces still seems flaky, but im in the process of purging wicd wpasupplicant etc.
One more thing noted:
root@beaglebone:~# apt-get remove xserver-xorg-core
Reading package lists… Done
Building dependency tree
Reading state information… Done
The following packages were automatically installed and are no longer required:
am335x-pru-package gyp javascript-common libjs-node-uuid libsoc2 libv8-3.14-dev libv8-3.14.5 node-abbrev node-ansi
node-archy node-async node-block-stream node-combined-stream node-cookie-jar node-delayed-stream node-forever-agent
node-form-data node-fstream node-fstream-ignore node-github-url-from-git node-glob node-graceful-fs node-gyp
node-inherits node-ini node-json-stringify-safe node-lockfile node-lru-cache node-mime node-minimatch node-mkdirp
node-mute-stream node-node-uuid node-nopt node-normalize-package-data node-npmlog node-once node-osenv node-qs
node-read node-read-package-json node-request node-retry node-rimraf node-semver node-sha node-sigmund node-slide
node-tar node-tunnel-agent node-which nodejs nodejs-dev nodejs-legacy npm wwwconfig-common xinput-calibrator
Use ‘apt-get autoremove’ to remove them.
The following packages will be REMOVED:
beaglebone xserver-xorg xserver-xorg-core xserver-xorg-input-all xserver-xorg-input-evdev
xserver-xorg-input-synaptics xserver-xorg-input-wacom xserver-xorg-video-fbdev xserver-xorg-video-modesetting
0 upgraded, 0 newly installed, 9 to remove and 0 not upgraded.
After this operation, 4127 kB disk space will be freed.
Do you want to continue [Y/n]? n
Abort.
Since when is X11 a requirement for ANY Nodejs package ? I’d have a assume these packages were somehow installed from within X that made such a requirement ? Either way, this is bad mojo.
Jason, Ok cloud9 seems to be functional, and I have to say this is pretty
awesome. Where can I find documentation on how you have this setup ?
Ok moving on to the other things for this image. First, there is no uEnv.txt
file *period*. We have bbb-uEnv.txt, and nfs-uEnv.txt. Neither one of these
will work, because I have the original uboot from an A5A. So, I follow
Roberts build from source instructions, so I know how to deal with this.
Someone else may not, but this is a testing image after all - I guess.
rename "bbb-uEnv.txt" as "uEnv.txt" and it'll boot with your old
"original uboot"..
I'm still poking around, but I disabled HDMI in uEnv.txt, and the image now
seems to be more responsive. networking interfaces still seems flaky, but im
in the process of purging wicd wpasupplicant etc.
wicd handles eth0, so re add it to /etc/network/interfaces
(boot time)
Regards,
It's the "beaglebone" meta-package. By removing xserver-xorg-core, you
also removed xserver-xorg-video-modesetting, which is specified in
"beaglebone". So everything else specified in "beaglebone" is now
marked as not directly installed (node-*/etc) but "not removed".
To fix this, i would have to change (1), to a list with every node-xy package..
1: https://github.com/RobertCNelson/omap-image-builder/blob/master/configs/bb.org-debian-stable.conf#L47
Yeah, patches welcome..
Regards,
wicd handles eth0, so re add it to /etc/network/interfaces
Already had, but something, I’m assuming wicd was causing problems. Routing was not being setup correctly. However, I was not 100% exactly what was causing these issues as setting hwdaddr as suggested in the interfaces file seemed like it could have been the cause as well.
apt-get purge wpasupplicant && apt-get autoremove --purge
Seemed to work just fine for me. But I do not need / use a wifi device.YET
So, I suppose I do not get the purpose of these testing images if it is not to let you, and the other maintainers know what we find wrong with them. I have no problem making something work for myself usually, and would not even have a problem creating patches. If I knew how.
Personally I would like to see a “minimal” console image without a lot of bloat, that has this very awesome looking cloud9 version working. But if that not your goal or the goal of beaglebone.org, then I suppose I could just get something working for myself and continue on my merry way. Again, I usually have no problems getting things I want done - done.
I will try to see if I can figure out your build script here, and see if I can make it so those of us who use autoremove do not have to worry about trashing our tools.
wicd handles eth0, so re add it to /etc/network/interfaces
Already had, but something, I'm assuming wicd was causing problems. Routing
was not being setup correctly. However, I was not 100% exactly what was
causing these issues as setting hwdaddr as suggested in the interfaces file
seemed like it could have been the cause as well.
try:
apt-get wicd* --purge
apt-get purge wpasupplicant && apt-get autoremove --purge
Seemed to work just fine for me. But I do not need / use a wifi device.*YET*
So, I suppose I do not get the purpose of these testing images if it is not
to let you, and the other maintainers know what we find wrong with them. I
have no problem making something work for myself usually, and would not even
have a problem creating patches. If I knew how.
No you get the purpose, i just don't know how to fix the bone101, and
the meta "beaglebone" package is one i don't really want to fix, this:
Robert, ok, cool. I’ve created packages before, and one recently for Nodejs. This is how my own 3.8.13-bone47 + barefs is so small with Nodejs + express + socket.io. I would not be above doing the same for this cloud9, and sharing the information in one way or another.
I’m still very newb with git, and several aspects of “embedded” Linux, Nodejs, etc. SO learning how to create a diff patch, etc . . . well let me just say it is something i want to learn, but need the time. I will look at all the relivent information you’ve given here, and see what I can come up with on my own. Then I’ll perhaps ask specific questions when / if I need to.
To this point:
Otherwise, anytime you remove something auto installed by “beaglebone”
it’ll show one of above as "The following packages were automatically
installed and are no longer required"
Yeah I get that, the “use autoremove” etc, but did not realize it was a package created by you all. I see that it was now. I know that dpkg, and perhaps apt too ( pretty sure aptitude could handle it as well ) can flag packages as required for whatever, thus leaving the packages safe from autoremove. However, I am still not 100% sure what cloud9 needs to run properly.
I do know that using APT as discussed previously will break cloud9. It’s still there looking all pretty, but the console does not work and the premeade project files are all gone.
Anyway, I’ll work on this a little at a time as I think this is something others will want too. I’m not sure if anyone else cares about what else is running on the BBB, but i do . . .I’ll have to toss together a NFS root test “jig” for this so it may not be right away.
I can see multiple ways to approach this, maybe it would be best to start from the console image, and install everything needed, manually, then giving you all the instructions needed to duplicate ? Or i could just do another blog post if there is not enough call to make a completely different image to host.
Again, I do not know who else cares enough about their BBB images to want this. Personally, I do not need this myself technically( I do think its pretty cool ), but I can see where it would / could come in handy for many people at work, or school, etc.
>> looks like the last push is missing the defconfig:
>>
http://builds.beagleboard.org/builders/runtests/builds/6/steps/shell/logs/stdio
>
> It's being really weird:
>
> http://builds.beagleboard.org/one_line_per_build
>
> Aug 20 21:22 b14c60a94f59...successruntests #5 Build successful
> Aug 20 21:20 20830c6f9585...failureruntests #4 Failed shell shell_1
>
> When i pushed the branch yesterday... It decided to build twice...
> The first build, took the first commit over the old branch, which
> failed as bb.org_defconfig is the last commit.. The second build
> completed correctly.
>
> Looks like build#6 is a rebuild of build#4
>
> can you add:
actallly can you add:
if [ -e arch/arm/configs/bb.org_defconfig ] ; then
config="bb.org_defconfig"
else
config="omap2plus_defconfig"
fi
Regards,
I'd prefer to generate builds that test that what we want to build, not
some other config or nothing at all. I tried adding a 2 minute stability
timer in hopes that would clear up it building the wrong thing.
I'm thinking the next step is publishing the build somewhere. Any
suggestion on how best to package it?
make ARCH=arm KBUILD_DEBARCH=armhf LOCALVERSION=-git<version>
CROSS_COMPILE=<bin>/arm-linux-gnueabihf- KDEB_PKGVERSION=1cross
deb-pkg
Then just copy the *.deb to an apache directory..
wget xyz.deb
sudo dpkg -i xyz.deb
sudo reboot
Regards,
I'd prefer to generate builds that test that what we want to build, not some
other config or nothing at all. I tried adding a 2 minute stability timer in
hopes that would clear up it building the wrong thing.
That worked:
http://builds.beagleboard.org/one_line_per_build
http://builds.beagleboard.org/builders/runtests/builds/12
Regards,
Robert, which build system are you using that makes it that “easy” to build deb’s ?
I’ve been using checkinstall . …
Mainline kernel has a "make deb-pkg".. I've been using that since
2.6.32 to populate rcn-ee.net/deb/ (each deb built in a chroot to
match it's target os)
Just recently, (after mainline decided where they were going to store
the *.dtb's) i added this patch to make sure they were also included.
https://github.com/RobertCNelson/stable-kernel/blob/master/patches/packaging/0002-deb-pkg-install-dtbs-in-linux-image-package.patch
Thus we now can generate a *.deb that contains everything we need to boot..
Regards,