<sveinse>
My use case is that I have to wipe my tmp and start a new build, but I'd like as much as possible of the current recipes into the sstate cache before the wipe
gioyik has quit [Ping timeout: 276 seconds]
gioyik has joined #yocto
gioyik has quit [Ping timeout: 276 seconds]
prabhakarlad has quit [Quit: Client closed]
nad has quit [Quit: Client closed]
gioyik has joined #yocto
gioyik has quit [Ping timeout: 276 seconds]
mattsm has quit [Read error: Connection reset by peer]
prabhakarlad has joined #yocto
mattsm has joined #yocto
Tokamak has quit [Ping timeout: 256 seconds]
Tokamak has joined #yocto
gioyik has joined #yocto
gioyik has quit [Ping timeout: 276 seconds]
Tokamak has quit [Read error: Connection reset by peer]
Tokamak has joined #yocto
mattsm has quit [Quit: Ping timeout (120 seconds)]
mattsm has joined #yocto
camus has quit [Ping timeout: 245 seconds]
camus has joined #yocto
goliath has quit [Quit: SIGSEGV]
RobertBerger has joined #yocto
rber|res has quit [Ping timeout: 260 seconds]
sakoman has quit [Quit: Leaving.]
prabhakarlad has quit [Quit: Client closed]
sakoman has joined #yocto
<kergoth>
sveinse: sstate, like bitbake, is task based, not recipe based, so that'd be unusual behavior to try to 'finish' a recipe as much as possible unless you're in an rm_work sort of operation. i can see why you'd want it, but i think you probably see why it wouldn't be default behavior :)
gioyik has joined #yocto
camus has quit [Quit: camus]
camus1 has joined #yocto
gioyik has quit [Ping timeout: 276 seconds]
camus1 has quit [Ping timeout: 250 seconds]
gioyik has joined #yocto
Vonter has joined #yocto
gioyik has quit [Ping timeout: 276 seconds]
gioyik has joined #yocto
sakoman has quit [Quit: Leaving.]
ilunev has quit [Ping timeout: 250 seconds]
gioyik has quit [Ping timeout: 276 seconds]
ilunev has joined #yocto
<kroon>
RP, kanavin, at least my image built without problems using BUILD_LDFLAGS += "-Wl,--build-id=none". will start a new build with a different TMPDIR
<kroon>
kergoth, sveinse, sometimes I run into the problem that the disk my TMPDIR is on gets full. if so, then I'd like to finish any big recipe-builds, but I can't start any clean tasks in order to free up space since no new tasks are allowed
<kroon>
kergoth, sveinse, would be nice if that was somehow allowed
ilunev has quit [Read error: Connection reset by peer]
ilunev has joined #yocto
alessioigor has joined #yocto
<kroon>
RP, kanavin, also, I think we need to test reproducilibity with building in two buildpaths that have different absolute length. so not {/usr/sysrootA,/usr/sysrootB}, but {/usr/sysroot/A,/usr/sysroot/the/brown/fox/jumped/B} or something like that
alessioigor has quit [Quit: alessioigor]
<kroon>
RP, kanavin, because even after removing the build-id, there is a null-section in the binaries that differ in size, which looks to be related to the absolute build paths
GNUmoon has quit [Ping timeout: 276 seconds]
<kroon>
RP, kanavin, yes, its .dynstr that differs in size, and that section contains the rpath strings
<kroon>
RP, could we just pass rpath = "$ORIGIN/../lib:$ORIGIN/../../lib" directly in BUILD_LDFLAGS ? that should fix the section lengths. ill try
vd has quit [Ping timeout: 256 seconds]
leon-anavi has joined #yocto
GNUmoon has joined #yocto
gioyik has joined #yocto
<kroon>
RP, then we could even keep the build-id
gioyik has quit [Ping timeout: 276 seconds]
lowfi has quit [Read error: Connection reset by peer]
rob_w has joined #yocto
tnovotny has joined #yocto
<kroon>
RP, i think using a constant "placeholder" in rpath instead of STAGING_LIBDIR_NATIVE and STAGING_BASE_LIBDIR_NATIVE will solve a lot of things
rfuentess has joined #yocto
<michalkotyla>
Hello, is there any confirmed and working way to copy files into the system image? I try with "install" inside do_install() and it set my files inside workdir/image. These files exist also in workdir/packages-split. Does someone have an idea what I should check next? Of course, I tried to clean the recipe with "cleanall" and rebuild it again - nothing is changed.
mckoan|away is now known as mckoan
zpfvo has joined #yocto
fleg has joined #yocto
olani has joined #yocto
rob_w has quit [Remote host closed the connection]
roussinm1 has quit [Quit: WeeChat 3.3-dev]
mvlad has joined #yocto
kayterina has joined #yocto
zpfvo has quit [Ping timeout: 250 seconds]
zpfvo has joined #yocto
zpfvo has quit [Ping timeout: 260 seconds]
zpfvo has joined #yocto
olani has quit []
<qschulz>
michalkotyla: oe-pkgdata-util find-path "/path/to/your/file/on/target" should return a package, add this package to your IMAGE_INSTALL in your recipe
mihai has joined #yocto
mattsm has quit [Quit: Ping timeout (120 seconds)]
mattsm has joined #yocto
zpfvo has quit [Ping timeout: 268 seconds]
zpfvo has joined #yocto
oberon has joined #yocto
<oberon>
I wanna add open sshd server to an image I create
<oberon>
to do so I added the following line to local.conf:
<oberon>
when I boot from the created image I can see sshd installed, but it isnt started (dont see it in pstree)
td has joined #yocto
<oberon>
if I run "systemctl start sshd" it starts
<oberon>
why do I need o do it manually ?
<oberon>
*to
<td>
Hello, is there any reason why "linux-yocto" recipe is called has a suffix "-yocto"? Is there any recommendation to have our custom linux recipes called "linux-foobar" instead of just "linux"?
zpfvo has quit [Ping timeout: 250 seconds]
<qschulz>
td: how would you be able to pick between linux recipes if all were named the same?
zpfvo has joined #yocto
oberon is now known as oberonc
<td>
if I only have single linux recipe that wouldn't be a problem
<qschulz>
td: well, you already have linux-yocto so it's already one :)
<qschulz>
you can name it linux, it's just that if there';s a name conflict, it is possible your recipe won't be taken
<td>
so there are no constraints from bb classes point of view, it's just to avoid conflict, am I right?
zpfvo has quit [Ping timeout: 250 seconds]
zpfvo has joined #yocto
tnovotny has quit [Quit: Leaving]
manuel1985 has joined #yocto
<qschulz>
td: there are constraints in naming convention followed by bitbake
<td>
could you elaborate on these convention in reference to linux?
<qschulz>
namely, lower case letters only, only dashes except between the name of the recipe and the version of the recipe
<qschulz>
i.e. some-recipe_1.2.3.bb
<qschulz>
perfect example being wpa_supplicant whose recipe filename is wpa-supplicant_xxx
<td>
both linux and linux-foo match the convention
<qschulz>
yup, they are thus valid
td38 has joined #yocto
zpfvo has quit [Ping timeout: 268 seconds]
zpfvo has joined #yocto
td38 has quit [Client Quit]
td has quit [Quit: Client closed]
<qschulz>
oberonc: Assuming it's your new nick from oberon, can you run bitbake openssh -e | grep "^SYSTEMD_AUTO_ENABLE" and tell us what's the output of that?
<qschulz>
oberonc: not verse with systemd sorry :/
<oberonc>
I accidentally deleted the contents of the build/tmp-glibc/deploy/images/intel-x86-64 folder
<oberonc>
now bitbake fails and says that bzImage in it doesnt exist
<qschulz>
you can remove everything except sstate-cache and downloads directories IIRC
<oberonc>
how do I recreate its content ?
zpfvo has quit [Ping timeout: 245 seconds]
zpfvo has joined #yocto
zpfvo has quit [Ping timeout: 245 seconds]
zpfvo has joined #yocto
<manuel1985>
oberonc: Would try a `bitbake -c clean <image-name>`
<manuel1985>
FYI, with `bitbake -c listtasks` you can get bitbake to list all tasks there are. There are a few related to cleaning.
<oberonc>
ok thanks
prabhakarlad has joined #yocto
mattsm has quit [Remote host closed the connection]
mattsm has joined #yocto
alessioigor has joined #yocto
alessioigor has quit [Quit: alessioigor]
alessioigor has joined #yocto
alessioigor has quit [Remote host closed the connection]
alessioigor has joined #yocto
alessioigor has quit [Quit: alessioigor]
zpfvo has quit [Ping timeout: 245 seconds]
zpfvo has joined #yocto
florian has joined #yocto
MauroAnjo has joined #yocto
goliath has joined #yocto
<MauroAnjo>
Hi! I'm currently trying to boot poky honister on a pi zero 2 W with dtoverlay=dwc2,dr_mode=peripheral, to use ethernet over USB. But on dmesg it states "dwc2 : Configuration mismatch. dr_mode forced to host"
<MauroAnjo>
so I can't up the interface usb0, only wifi
Guest45 has quit [Quit: Client closed]
xmn has joined #yocto
florian_kc has joined #yocto
kayterina has quit [Ping timeout: 268 seconds]
<oberonc>
the sshd that doesnt start problem I mentioned earlier
<oberonc>
I was mistaken
kayterina has joined #yocto
<oberonc>
it doesnt start manually either with the error:
<oberonc>
sshd: no hostkeys available - exiting
<oberonc>
as if it doesnt run "/usr/bin/ssh-keygen -A"
<oberonc>
how so ?
pregn has joined #yocto
<pregn>
hello
<pregn>
just a simple question, I have a SRC_URI in my project that needs updating: I need to fetch a precise commit number do i use SRCREV? and how?
<rburton>
yes, assuming it fetches using it, just put the SHA you want in SRCREV
<rburton>
"uses git"
zpfvo has quit [Ping timeout: 250 seconds]
<pregn>
oh fab thanks
zpfvo has joined #yocto
zpfvo has quit [Ping timeout: 260 seconds]
zpfvo has joined #yocto
<qschulz>
oberonc: is ssh-keygen present on the target?
<pregn>
rburton: the error I get now is due to git ls-remote not finding the commit number I need in the branch, only the latest revision of all the branches is lilsted by this command
<rburton>
if the sha is on a branch then you might need to specify the branch in the SRC_URI like ;branch=foobar, as the fetcher will validate the sha is on the branch you say it is
<rburton>
easier to help if you can share the src-uri and sha you're trying to use, assuming its publically available
<pregn>
that is using SRCREV="<commit number>" and SRC_URI="gitsm://git@bitbucket.org/whatever/project.git;protocol=ssh;branch=patrick-develop"
<oberonc>
ssh-keygen is present
<oberonc>
I boot from the .iso image and it is read only
<qschulz>
well that might explain it
<oberonc>
maybe it doesnt run ssh-keygen after booting
<qschulz>
ssh-keygen will try to write the key
<rburton>
pregn: assuming the sha in SRCREV is on the branch you specify, it should work. if not, paste the actual error
<qschulz>
but if it's RO, it won't manage to
<oberonc>
but shouldnt it run ssh-keygen during creation of the image ?
<oberonc>
so it would boot from an iso image that is RO but which already contains the host keys ..
<qschulz>
oberonc: I doubt it does it by default since it'd mean all devices on which you flash your device would have the same keys
<rburton>
see eg the ssh-pregen-hostkeys recipe for how to build your own keys at image time, not boot time
<qschulz>
rburton: to the rescue :)
<rburton>
identical keys for every deployment is not a great idea for obvious reasons
<oberonc>
hmm
<rburton>
don't use that recipe as-is, generate your own keys
<oberonc>
that was a command to install from a booted iso image to local harddrive.. what was it ?
<oberonc>
*there
<rburton>
i guess the server should handle read-only and no keys, and generate keys in /run
<oberonc>
had this command available:
<oberonc>
tgt=/dev/sda /sbin/reset_media
<pregn>
MyProject-0.1-r0 do_fetch: Fetcher failure: Unable to resolve '1b984ee' in upstream git repository in git ls-remote output for bitbucket.org/Whatever/MyProject.git
<pregn>
rburton: the command in question just lists branches and their latest commit SHA
<rburton>
presumably this is a private repo so you can't share real urls
<pregn>
rburton: correct
<rburton>
it should work with short shas but maybe try using the full sha
<pregn>
will try with full SHA
<pregn>
one moment
<rburton>
good to be precise anyway, you don't want to debug problems where two commits match the shortened name
MauroAnjo has quit [Quit: Client closed]
<pregn>
rburton: Correct, seems to work with the long HASH
<pregn>
now I have to regress the bb file as well, should be straight forward
<pregn>
rburton: thank you so mch for your help
<rburton>
np
leon-anavi has quit [Read error: Connection reset by peer]
Guest11 has joined #yocto
<Guest11>
Hey team!
leon-anavi has joined #yocto
<Guest11>
I'm completely new to this.
<Guest11>
We are using Jenkins for our BitBake build
<rburton>
there was work on the servers last night, maybe something just broke
<rburton>
'The migration is complete. git.yoctoproject.org now serves from a pair of load balanced mirrors. push.yoctoproject.org is on a new dedicated secure host.'
zpfvo has quit [Ping timeout: 250 seconds]
<rburton>
ah after about a minute of hanging, its fetching at 2mib/s, so slowly
<kroon>
I'm playing around with the native rpaths, and suddently python3-native creates <...>-gnu_failed.so files. Anyone got a clue what that is about ?
<Saur>
First time I tried to clone poky, I got 12.00 KiB/s. Now I get 502...
zpfvo has joined #yocto
<Saur>
Error 502, that is.
<RP>
I'm seeing intermittent 502s as well
<Guest11>
Oh okay. Bad times. We were meant to release today
<Guest11>
Is there any way I can help?
<Guest11>
Hopefully someone with the right context/perms and authority is in this chat 🤞
mckoan is now known as mckoan|away
<RP>
Guest11: the person we need is a asleep at the moment, it is 5am for them
vijay has joined #yocto
<Guest11>
Oh okay
<Guest11>
I can wait :)
xmn has quit [Quit: ZZZzzz…]
<Saur>
Guest11: But now you have learnt that it is a good idea to have a local mirror if you want to make releases. ;)
gourve_l has quit [Quit: WeeChat 2.3]
jwillikers has joined #yocto
mattsm has quit [Remote host closed the connection]
mattsm has joined #yocto
<oberonc>
how do I get bitbake to print possible targets ?
jwillikers has quit [Remote host closed the connection]
<qschulz>
find -type d -name "machine" -exec ls -l {} \;
<oberonc>
that is an answer to my question ?
<qschulz>
one way yes
<qschulz>
all machines are in conf/machine of any layer that is in build/conf/bblayers.con
<qschulz>
f
<qschulz>
actually probably ls -l {}/*.conf \;
jwillikers has joined #yocto
<qschulz>
since only files ending with .conf in conf/machine/ are valid machines
<qschulz>
(well technically, it's based on BBPATH, but I assume the majority of layers aren't messing with that one)
jwillikers has quit [Ping timeout: 240 seconds]
oberonc has quit [Quit: Client closed]
jwillikers_ has joined #yocto
zpfvo has quit [Ping timeout: 260 seconds]
zpfvo has joined #yocto
<rburton>
a list of recipes is available from bitbake-layers
<rburton>
but i'm too late to tell oberonc
<qschulz>
rburton: yeah I don't know if they were looking for targets like.. machine or bitbake targets (recipes)
Guest39 has joined #yocto
Guest39 has quit [Client Quit]
zpfvo has quit [Ping timeout: 245 seconds]
zpfvo has joined #yocto
gioyik has joined #yocto
RobertBerger has quit [Ping timeout: 256 seconds]
jwillikers_ has quit [Remote host closed the connection]
rber|res has joined #yocto
GNUmoon has quit [Remote host closed the connection]
GNUmoon has joined #yocto
<jaskij[m]>
are there any docs on packaging Go binaries newer than 2018?
<jaskij[m]>
are you sure that commit/tag is actually in the default (master) branch?
<jaskij[m]>
also: IIRC there's Beaglebord BSPs freely available
<jaskij[m]>
ah, those are for Beaglebone only it seems
lucaceresoli has joined #yocto
<jaskij[m]>
sinic: TL;DR, you might need to specify `branch=something` in that `SRC_URI` `git://` entry
<sinic[m]>
jaskij[m]: no, but how do i check that / get the branch for that tag?
<sinic[m]>
jaskij[m]: yes, they are, but none of them are working really well with the pocketbeagle and PRU. At least not out of the box. And i wanted to do some basics :)
<sinic[m]>
* do some yocto basics :)
<jaskij[m]>
sinic[m]: `git branch -a --contains commit_hash`
<sinic[m]>
`* (HEAD losgelöst bei 50301b35d336)`
<qschulz>
sinic[m]: if it's a tag, just provide ;tag= to your SRC_URI
<sinic[m]>
It seems there are no branches containing these tags ...
<sinic[m]>
i did, but i get the exact same error
<qschulz>
I'm wondering if you need SRCREV if you provide a tag
<qschulz>
I'd say no
<sinic[m]>
no, it is commented out
<smurray>
it's much better to use the commit sha in SRCREV that's tagged instead of a tag, since tags can be changed
<smurray>
err, commit id that's tagged, -ENEEDCOFFEE
<sinic[m]>
smurray: i also tried that, but it has the same effect (fetch error) as when i use the `;tag=` annotation
<smurray>
sinic[m]: if the commit isn't on master, try specifying ;branch=
<zeddii>
if it isn't on any branch, then that's probably not the commit ID for you :) even if it was tagged, since that shows some seriously bad git repo maintenance. but you can put nobranch=1
<sinic[m]>
zeddii: It's the official beagleboard linux kernel, so not in my hands to change that 😄
<AlejandroExojo[m>
I am very confused about how one is supposed to lay out in disk the layers and the build directory. The reason is that I want to have the tree under version control. But stuff like bblayers.conf is under local configuration, right? So how does one share with a team a "distro", including the wanted layers?
<AlejandroExojo[m>
The shared projects online that I see (e.g. poky) use a script that copies a template to the conf directory. But if one adds a layer, e.g. via bitbake-layers, that doesn't edit the template, edits the generated file. So what am I missing?
zpfvo has quit [Ping timeout: 240 seconds]
zpfvo has joined #yocto
rob_w has quit [Remote host closed the connection]
<zeddii>
haha. cut and paste rushing on the package name :P thanks for the fixups.
zpfvo has quit [Ping timeout: 250 seconds]
zpfvo has joined #yocto
Tyaku has quit [Quit: Lost terminal]
zpfvo has quit [Ping timeout: 245 seconds]
zpfvo has joined #yocto
RobertBerger has quit [Ping timeout: 250 seconds]
rber|res has joined #yocto
sakoman has joined #yocto
kayterina has quit [Read error: Connection reset by peer]
doquiros[m] has quit [Quit: You have been kicked for being idle]
zpfvo has quit [Ping timeout: 260 seconds]
zpfvo has joined #yocto
kroon_ has joined #yocto
kroon has quit [Ping timeout: 245 seconds]
gioyik has quit [Ping timeout: 276 seconds]
<rburton>
sinic[m]: if the SHA you want isn't actually on a branch then set nobranch=1
Guest55 has joined #yocto
<sinic[m]>
But can you tell me, why this is necessary?
<sinic[m]>
rburton: yeah, I did and it works now, thanks 😊
<Guest55>
Dear all, has anyone experienced extremely slow downloads from downloads.yoctoproject.org/mirror/sources ?
<Guest55>
I'm getting like ~20kbit/s from a 200mbit/s broadband connection
gioyik has joined #yocto
barometz has joined #yocto
<rburton>
sinic[m]: because the fetcher sanity checks that the sha you asked for is on the right branch
<rburton>
just a bit of sanity checking really
<qschulz>
AlejandroExojo[m: kas on github allows you to configure the layers you need where to fetch them and some other stuff
<qschulz>
I'll start using in the next few weeks for CI but I don't know enough yet, but that's one of the tools people are using
vd has joined #yocto
kiran_ has joined #yocto
rfuentess has quit [Remote host closed the connection]
<Guest11>
RP is the person who looks after the git mirror around?
<RP>
Guest11: we need halstead
gioyik has quit [Remote host closed the connection]
<alex88>
is it possible to inspect the final PACKAGECONFIG of a recipe? bitbake -e doesn't show that (because it's only relative within a recipe I think)
<RP>
alex88: bitbake <recipe> -e should
<RP>
or bitbake-getvar
Epictek[m] has joined #yocto
<alex88>
oh thanks RP, didn't try with the recipe name
<alex88>
why would PACKAGECONFIG:append = "networkd resolved polkit" add the values properly and PACKAGECONFIG += "networkd resolved polkit" result in having only those 3? (systemd in this case)
gioyik has joined #yocto
<Epictek[m]>
<Guest55> "Dear all, has anyone experienced..." <- Yeah, somethings broke :)
<alex88>
rfs613, so better to stick with :append when possible?
<dj>
it takes a lot of bandwidth having to git clone poky over and over for each repos or testing clean repos
<Guest11>
Guest55 thanks. I posted about it about 5 hours ago but halstead was still sleeping. If I can assist with anything (testing/pinging/verifying) let me know
<rfs613>
alex88: I'm not expert, but I think :append is the preferred way
<rburton>
dj: why would you need to clone it repeatedly?
<rburton>
also what package manager would it be in
<dj>
apt? (ubuntu)
<rburton>
that would make no sense at all
<dj>
and i clone it because i git clone a repo that git clones poky, and I want to test if a clean repo is successful in a build
<Guest55>
Guest11: I wish I knew of another mirror, to at least bridge the time until the official one works again
<rburton>
so an apt package wouldn't test that then
<rfs613>
alex88: later on the same page I linked, in section 3.4.2, it says "If VARNAME is going to be set, it needs to be set before the inherit statement is parsed"
<Epictek[m]>
<Guest55> "Epictek: so issue is known and..." <- Know by the pocky devs? No idea. Known by myself and my colleagues? Yes :D
zpfvo has quit [Ping timeout: 268 seconds]
<dj>
i imagine i wouldn't have to download as much data if there was a yocto/poky package
<rburton>
dj: look up git reference directories, to tell git to look in a local directory as a download cache
<Guest55>
Epictek[m]: (y)
<Guest55>
Epictek[m]: do you know of a public available mirror, preferably in europe, to bridge the time?
<Guest11>
Epictek[m], RP mentioned that halstead is the person we need? Also what Guest55 said. Also apologies for all the tags
<RP>
Guest55: which repo are you looking for? poky?
<rburton>
the yocto git server appears to be back again now fwiw
<Saur>
For what it's worth, I get 7-12kB/s at work (in Sweden). However, if I clone from a server at the university that I happen to have access to, then I get 8 MB/s...
<RP>
Saur: can you see which IPs those resolve to?
<Saur>
RP: You mean for git.yoctoproject.org?
<RP>
Saur: yes, I'm just wondering if that was the same server or two different ones
<rburton>
i got a fast fetch and git.yp is 198.145.29.87
<Saur>
Both resolv to 198.145.29.87
<RP>
ok. I don't know how the load balancing is being done
<alex88>
rfs613, forgot to mention, I did use both ways in a bbappend
<alex88>
so maybe in the main recipe it was set in a different way
zpfvo has joined #yocto
<alex88>
probably because the main recipe uses PACKAGECONFIG ??=
<alex88>
so += sets the variable and then ??= doesn't set the default
<AlejandroExojo[m>
qshulz: thank you. It doesn't seem (to me) like the right solution (adding yet another configuration file to learn and another tool to install), but I'll consider it. I don't understand how this is not built in.
gioyik has quit [Quit: WeeChat 3.1]
<rfs613>
hmm, since github has deprecated unauthenticated git protocol, I see a lot of SRC_URL having "protocol=https" appended. Is there a reason for this, rather than changing the prefix from git:// to https:// ?
zpfvo has quit [Quit: Leaving.]
<rburton>
the prefix is what high level protocol to use, so 'git'
<rburton>
not what underlying protocol that uses
<rburton>
if you put https: it would try to wget the git repo
<rburton>
which wouldn't work
<rburton>
for example there's also a s3:// top level protocol. yes, that's https, but it's done specially.
<rfs613>
right, thanks - so it uses https but doesn't just do a single GET of the SRC_URL, it does something git-ish..
<rburton>
well, it calls git
<rburton>
its a git clone, with various options, then checks out the sha you want
<rfs613>
so I guess it replaces the git:// with https:// when passing the URL to git
<rfs613>
but we need the git:// to trigger the git-fetcher to call git
<rburton>
yeah, the protocol parameter is the actual protocol to give to git
<Guest11>
RP do you know if you are using Github Load Balancer (GLB) for the repo (https://github.com/github/glb-director#:~:text=The%20GitHub%20Load%20Balancer%20(GLB,all%20traffic%20from%20GitHub's%20datacenters.)
<rburton>
we're not using github for hosting
<RP>
Guest11: it isn't using github as rburton says
<Guest11>
rburton This is an OSS load balancer which you can deploy for your own infra?
<Guest11>
I'm just trying to understand how this load balancing is happening. Maybe I can specify another IP
<RP>
Guest11: Yocto Project is a Linux Foundation project and it is using the same git hosting infrastructure as the kernel git repos. We don't know more, we need the sysadmin who look after it
<Guest11>
RP thanks! halstead hasn't responded perhaps he's looking into the issue
<Guest55>
I wonder if the ISP is the problem... if I download a source from http://downloads.yoctoproject.org using my browser, with ISP X, download is fast
<Guest11>
In my case: local (London,UK) + AWS (Dublin, Ireland) - both slow
<dj>
<rburton> dj: look up git reference directories, to tell git to look in a local directory as a download cache
<dj>
are you talking about git submodules?
<dj>
because i'm not finding anything for "git reference directories" on google
<halstead>
Reading through the back scroll. It appears that all the connections to downloads.yoctoproject.org and git.yoctoproject.org from the UK are extremely slow and have been for hours.
<rburton>
halstead: not all, my machine in cambridge is fast
<Guest11>
halstead yes 👍
<dj>
thanks will take a look
<Guest55>
halstead: seems like a routing problem, it is ISP dependent, I am in Berlin Germany, with Telekom downloads are fine, with Vodafone, extremely slow
<halstead>
Thank you Guest55 . That's helpful.
<Guest55>
Currently 4 running tasks (471 of 6317) 7% |############# |
<Guest55>
while with Telekom I reach normal speeds
<halstead>
Guest55: Could you run a traceroute to git.yoctoproject.org from each connection and email to mhalstead@linuxfoundation.org? It would be a great help.
<dj>
rburton, would you recommend to git clone to ~/poky and the just use git clone --reference against it for projects?
<halstead>
Guest11: If you could email a traceroute as well that would be useful.
<dj>
and then just use git clone*
<rburton>
dj: if you find yourself frequently cloning the same repo then yeah
<dj>
thanks
<rburton>
our CI uses a persistant reference repo as it clones from scratch every build
<rburton>
locally, my git clone was done once a year ago
<Guest55>
halstead: one sec, sure
pregn has quit [Quit: Client closed]
<Guest11>
halstead I'll send you my traceroutes as well?
<Guest11>
Sending now
<halstead>
Thank you Guest11 .
kroon_ has quit [Remote host closed the connection]
<Guest55>
halstead: sent
<halstead>
Thank you.
florian has quit [Quit: Ex-Chat]
kroon_ has joined #yocto
Guest12 has quit [Quit: Client closed]
florian_kc has quit [Ping timeout: 260 seconds]
<Guest55>
I'm seeing normal speeds again
<halstead>
Guest11: Thanks for the traceroutes. Both of those are slow connections?
<Guest55>
wooohooo
<halstead>
Guest55: Good to know. I've raised the issue and provided your traceroutes. I'm not sure what's changed yet though.
prabhakarlad has quit [Ping timeout: 256 seconds]
<Guest55>
halstead: thanks for the help! <3
<Guest11>
halstead yes both
<Guest11>
Guest55 checking on my side
* halstead
nods.
<Guest11>
I'm seeing an improvement (539.00 KiB/s) but was hoping for MiB/s :D
<Guest55>
Guest11: brexit:/
<Guest55>
:D
<Guest11>
Guest55 that would have hurt if I was British lol
<Guest55>
:D
prabhakarlad has joined #yocto
<Guest11>
We use Jenkins and we can't seem to change the global git timeout config
<Guest11>
So we're stuck with the default 10 minutes
<zeddii>
so many guests
<Guest11>
It's just me and Guest55 being noisy
kroon has joined #yocto
kroon_ has quit [Ping timeout: 260 seconds]
kroon_ has joined #yocto
Guest11 has quit [Quit: Connection closed]
GNUmoon has quit [Remote host closed the connection]
xmn has quit [Quit: ZZZzzz…]
Guest5735 has quit [Ping timeout: 256 seconds]
kroon has quit [Ping timeout: 245 seconds]
Guest1125 has joined #yocto
<Guest1125>
Hello it's Guest11 my laptop went on stand by and I lot the connection :(
<Guest1125>
So I lost the chat history
davidinux has quit [Quit: WeeChat 2.8]
Vonter has quit [Ping timeout: 260 seconds]
Guest55 has quit [Quit: Client closed]
<Guest1125>
My fellow Guest55 left :/
inisider has joined #yocto
<AlejandroExojo[m>
You can set a name with /nick newname, that way it'll be less confusing. And you can access the chat history with the URLs on the topic
Guest1125 is now known as Mickey
Mickey is now known as Guest541
Guest541 is now known as mickeypash
kroon_ has quit [Quit: Leaving]
<inisider>
hello. I would like to ask how can I disable just one service using SYSTEMD_AUTO_ENABLE. I have SYSTEMD_PACKAGES = "${PN}-xen-watchdog ${PN}-xencommons ${PN}-xendomains" where SYSTEMD_SERVICE_${PN}-xen-watchdog = "xen-watchdog.service". Will SYSTEMD_AUTO_ENABLE_{PN}-xendomains="disable" work?
marc1 has quit [Ping timeout: 250 seconds]
GNUmoon has joined #yocto
<dj>
is there an environment variable to see where bitbake is downloading?
<dj>
because i'm apparently fetching something that I should already have I believe
GNUmoon has quit [Ping timeout: 276 seconds]
inisider has quit [Quit: Client closed]
<dj>
unless symlinks are breaking between docker & my host machine sharing the same ~/bitbake
marc1 has joined #yocto
GNUmoon has joined #yocto
manuel1985 has quit [Quit: Leaving]
kroon has joined #yocto
florian_kc has joined #yocto
Guest57 has joined #yocto
<rburton>
dj: it downloads into downloads/ under your build directory by default
<rburton>
dj: so if your build tree is in a directory that gets thrown away by docker, yes you'll be redownloading
<rburton>
stuff like the kas container or crops handles this and puts them outside of the docker
<rburton>
(for the exact path, see DL_DIR in your local.conf)
davidinux has joined #yocto
<dj>
so I should be setting my DL_DIR to ~/bitbake/downloads then?
<dj>
because I am throwing about my build tree to test PRs
<dj>
throwing away*
<rburton>
definitely put DL_DIR and SSTATE_DIR to somewhere that isn't thrown away
<rburton>
RP: just noticed the repo list on git.yp has a few new layers uncategorised, so they appear at the top
<rburton>
RP: where is that defined?
<rburton>
dj: is our CI, note how it sets DL_DIR and SSTATE_DIR to a persistent directory but does builds inside containers that are thrown away every run
<RP>
rburton: it is meant to be from the cgit config but I think halstead or I used to have to tweak from the commandline
<RP>
halstead: did that change now?
<halstead>
RP: rburton: it lives in a special ref now. Directions are https://korg.docs.kernel.org/cgit-meta-data.html. I added the data to all repos. Looks like it's being wiped out for a few of them.
<halstead>
All of these are mirrored from elsewhere. I need to get the cgit config on the upstream mirror.
<RP>
halstead: that explains things and nice we can set that now :)
leon-anavi has quit [Quit: Leaving]
<halstead>
RP: In the meantime I'll run a little cronjob to keep them sorted.
lucaceresoli has joined #yocto
Guest57 has quit [Ping timeout: 256 seconds]
GNUmoon has quit [Ping timeout: 276 seconds]
<halstead>
Those are getting corrected automatically now.
<RP>
halstead: thanks!
xmn has joined #yocto
kiran_ has quit [Ping timeout: 260 seconds]
florian_kc has quit [Ping timeout: 268 seconds]
GNUmoon has joined #yocto
FredO has joined #yocto
vijay has quit [Ping timeout: 260 seconds]
mvlad has quit [Remote host closed the connection]
<kroon>
RP, are you ok with adding reproducible tests for native recipes ? For instance openssl-native has patches for reproduce-ability but they are not functioning correctly on master any more
<RP>
kroon: as we have things which are reproducible, yes
<kroon>
RP, just saying it might make upgrading recipes more painful
<RP>
kroon: we probably can't take really invasive changes so it will really depend what kind of things we're looking at
<kroon>
RP, *nod*
lucaceresoli has quit [Ping timeout: 260 seconds]
<kroon>
RP, well at least I think there are things we can improve, with changes that shouldn't be too invasive
florian_kc has joined #yocto
kroon has quit [Quit: Leaving]
florian_kc has quit [Ping timeout: 260 seconds]
xantoz has quit [Ping timeout: 250 seconds]
<FredO>
Hi, when building in two different build directory (thats the only difference) using the same SSTATE_DIR; the second build hit only 50% from the sstate; I probably have something poluting the sstate with a PATH. Do you have any tricks to find which variables I should vardeps.../BB_HASHBASE_WHITELIST ?
<RP>
FredO: did you share a hashequiv server between them?
<RP>
if using hashequiv, then need the same hash database for sstate reuse
<FredO>
BB_HASHSERVE = "127.0.0.1:9082"
<FredO>
BB_SIGNATURE_HANDLER = "OEEquivHash"
<FredO>
that ?
<FredO>
RP: by hashdatabase you mean the one in my hash server right ?
<RP>
FredO: yes, as long as they share the same hash server
<FredO>
they do
Ad0 has quit [Ping timeout: 250 seconds]
<RP>
FredO: if you have the two builds and you have the different sig files, bitbake-diffsigs can show you the differences
<RP>
FredO: aim for the earliest difference in the dependency chain you can spot