maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
tlj_ has joined #openwrt-devel
tlj has quit [Remote host closed the connection]
maxmadzz has joined #openwrt-devel
<soxrok2212>
hmmm whenever i "echo out > /sys/class/gpio/gpio497/direction" my whole device crashes completely
<soxrok2212>
on xiaomi fw, not openwrt
maxmadzz has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
schwicht has joined #openwrt-devel
schwicht has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
lmore377 has joined #openwrt-devel
MaxSoniX has joined #openwrt-devel
rua has joined #openwrt-devel
Tapper has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
cbeznea has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
guidosarducci_ has joined #openwrt-devel
<PaulFertser>
soxrok2212: that likely means this gpio is supposed to be used as input and it's attached to some other external output. By forcing both to be outputs you create uncontrolled current flow.
MaxSoniX has quit [Ping timeout: 480 seconds]
MaxSoniX has joined #openwrt-devel
Tapper has quit [Ping timeout: 480 seconds]
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
rua has quit [Quit: Leaving.]
matoro has joined #openwrt-devel
rua has joined #openwrt-devel
<rsalvaterra>
Wow. Surprised/happy/relieved Yu Zhao himself stepped in to help with the MGLRU backport…!
Tapper has joined #openwrt-devel
<rsalvaterra>
I've rarely seen such support for kernel features, especially when it comes to backporting.
<jow>
hmm, I received reports that chrome 105 seems to have issues with LuCI over https (uhttpd) as shipped in 22.03
<jow>
I am very tempted to drop https by default again
<jow>
it achived nothing so far except regressions
<ynezz>
I've Chromium 106 here, and I'm not able to trigger this, bummer
borek has quit [Quit: borek]
maxmadzz has quit [Ping timeout: 480 seconds]
<jow>
ynezz: same here
<jow>
but could also be https on slow devices
<ynezz>
my zyxel,gs1900-8 is certainly not fastest :P
<ynezz>
Redmi Router AX6S is mt7622 so 64bit ARM
<ynezz>
might be related
<ynezz>
so maybe too fast
<ynezz>
puting jokes aside, maybe x86/64 qemu could exhibit this issue
guifipedro_ has joined #openwrt-devel
cbeznea1 has joined #openwrt-devel
maxmadzz has joined #openwrt-devel
cbeznea has quit [Ping timeout: 480 seconds]
Larhzu has joined #openwrt-devel
Borromini has joined #openwrt-devel
<\x>
I wonder how much help that MGLRU thing will do with 128MB ath10k devices
<\x>
Ive tried compiling it for a newifi 3 d2, it works but as you see that thing has 512MB of ram so I dont feel the differences hehe
<rsalvaterra>
\x: Actually, memory-constrained devices are set to benefit the most.
<\x>
>inb4 its too good that it makes 4/32 viable again
<Borromini>
=)
danitool has joined #openwrt-devel
<rsalvaterra>
\x: They're still perfectly viable, just not as routers. :P
<\x>
isnt some of those realtek switches come with specs like those?
<\x>
and thats perfectly fine for those hehe, after all, I cant think of anything that you want to add on a switch anyway, maybe luci-app-wol but thats about it
<rsalvaterra>
I was thinking more of paperweight…
* rsalvaterra
runs
robimarko has joined #openwrt-devel
maxmadzz has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
<Borromini>
\x: most have 128 MiB RAM AFAIK. And they don't have memory hogs like ath10k, so they are mostly fine.
<Borromini>
not running outdated vendor firmware on your hardware is one thing...
<Borromini>
turning an unmanaged switch into a managed one, is another
<jow>
now that we have masq6 "for free" with nftables, I wonder if we should an ipv6 masquerade rule for the ula range by default
<jow>
*add an
<jow>
it will have no impact on "proper" ipv6 setups
<jow>
but it would make local hosts having only ula addresses be able to reach the ipv6 internet
rua has quit [Ping timeout: 480 seconds]
<zorun>
jow: nat for ipv6? heretic! :)
<\x>
I think itll be fine, even here in my country, only two isps does proper ipv6-pd, others just get a single ipv6 address, theen theres those people who dualnats from their modems, it will atleast allow those to reach ipv6
<\x>
though even on those people that dulnats from their modem they can just use relay mode
<Borromini>
jow: sounds reasonable to me honestly
<jow>
okay, I'll think some more about it
<rsalvaterra>
zorun: When you have ISPs changing prefixes at will just because they can (i.e., to justify selling you fixed IP addresses), you need some form of NAT.
cbeznea has joined #openwrt-devel
<f00b4r0>
jow: I'm afraid your answer confuses me even more. What do you call "logical interface"? :)
<ynezz>
aparcar[m]: seems like a free lunch is over on gitlab.com CI, we're on 70% already in the current month 281/400 minutes
<jow>
f00b4r0: config interface lan; config interface wan
<aparcar[m]>
yea that's silly
<jow>
in /e/c/network
<aparcar[m]>
ynezz: I'd go for github until we have our own infra in some future
<aparcar[m]>
I don't see spinning up a gitlab runner what do you think
<f00b4r0>
jow: ok so the things that should really be called "networks" then. Now I get it. So the "virtual dynamic interface" should really be called a "virtual dynamic network" maybe?
<jow>
f00b4r0: "a dynamic virtual interface" is the same, except that it does not exist in /e/c/network but purely inside netifd's state. Protocol handlers spawn such "logical interfaces" by pushing the desired configuration to netifd via ubus, which then performes interface configuration as if it has been declared in /etc/config/network
<jow>
yes, correctly
<f00b4r0>
OK. I never completely understood why these extra interfaces/network were necessary. The fact that they don't belong to the "parent" firewall zone is also somewhat confusing when looking at Luci
<jow>
what luci calls "interfaces" are "logical uci networks"
<jow>
what luci calls "devices" some people call "interfaces" (kernel calls them netdevs)
cbeznea1 has quit [Ping timeout: 480 seconds]
<jow>
f00b4r0: they do belong to the parent firewall zone, but since this information is not exposed in ubus, luci does not know it / cannot show it
<f00b4r0>
yeah now it makes sense. I think "devices" is perfectly acceptable; especially since these show up in ip dev <whatever>
<f00b4r0>
ah!
<jow>
f00b4r0: or rather there's no information in the ifstatus of such virtual network that allows relating them to their parent network
<f00b4r0>
ok
<jow>
ideally I would assign them to th same zone and even show them indented next to their parent network
<jow>
but right now the crucial information is missing
<f00b4r0>
i see
rua has joined #openwrt-devel
<ynezz>
aparcar[m]: it's not for runners, it's CI minutes, I've just changed daily scheduled build check pipelines to monthly for several projects, but left docker as it is and ustream-ssl set to weekly
<f00b4r0>
eww. I see phase2 still uses rsync --checksum
<ynezz>
aparcar[m]: and added additional 5k minutes, we could probably as well try to apply for some FOSS tier, where they offer 50k minutes
<ynezz>
aparcar[m]: I assume, that other CI platforms would follow soon with cutting their free tiers as well, the costs are simply increasing a lot
Tapper has quit [Ping timeout: 480 seconds]
<ynezz>
hetzner.de just increased per server bill by 5EUR/month, due to electricity costs
<zorun>
rsalvaterra: no you don't. simply propagate the new prefix to clients through RA, it's designed for this
<aparcar[m]>
ynezz: I think some time ago people were against premium stuff from gitlab.com. Aren't their minutes calculated based on worker usage and not for actual runtime? Meaning if we have our own runner we have infinite minutes?
<rsalvaterra>
zorun: Hm… I wasn't aware of that, but I trust you, since I don't have much experience with IPv6 yet.
<zorun>
rsalvaterra: I must say I dislike dynamic prefixes (what kind of ISP would have such a bad idea?), but I dislike IPv6 NAT even more ;)
<rsalvaterra>
zorun: We've all been sold that IPv6 would spell the end of NAT… :)
<rsalvaterra>
Anyway, Portuguese ISPs which provide IPv6 do this, unfortunately.
<Borromini>
how does one handle dynamic prefixes with static IPv6 addresses? is there a way to get a 'WAN' IP with the same suffix?
<jow>
rsalvaterra: *properly implemented* IPv6 would end NAT
<robimarko>
rsalvaterra: Well, there is no reason for NAT in IPv6
<Borromini>
the hints in /etc/config/dhcp don't seem to work too well for me
<robimarko>
Only stupid ISP-s
<rsalvaterra>
robimarko: If they want to sell you fixed addresses, yes, there's a (greedy) reason for NAT.
<jow>
not handing out enough addresses or making the handed out addresses unusable will of course not help removing the need for NAT
<robimarko>
rsalvaterra: Well, I blame RIPE and company for this
<robimarko>
As they have setup a whole system that suits ISP-s perfectly
<robimarko>
FFS, selling IPv6 adresses in 2022
<rsalvaterra>
Don't attribute to incompetence that which can be explained by greed. :P
<robimarko>
IPS-s like mine that dont offer IPv6 at all are even better
<ynezz>
aparcar[m]: yes, "Jobs on specific runners are not affected by the quota of CI/CD minutes."
<ynezz>
aparcar[m]: do we need to build those docker containers on daily basis?
<aparcar[m]>
ynezz: no let's just stop snapshot containers for now. We could just build them in the github CI on the fly and cache them for 24 hours
<aparcar[m]>
I'll try to look into that but disable them for now if you can
<ynezz>
weekly should be good enough
<ynezz>
lets see if anyone even notices this :p
rsalvaterra has quit []
rsalvaterra has joined #openwrt-devel
<f00b4r0>
ynezz: any interest in implementing running snapshot and/or release builders on a daily/weekly basis instead of continuous, to save on resources?
<f00b4r0>
(while I've got my hands on the buildbot files :)
<robimarko>
I can see released versions getting built like that, but not snapshots
<ynezz>
f00b4r0: as discussed in private, you're not going to save anything, until we refactor the buildbot setup dramatically
<ynezz>
f00b4r0: you can't shutdown the servers between builds
<f00b4r0>
ynezz: I'm talking about saving cpu cycles
<robimarko>
If they are burning energy anyway, whats the point?
<f00b4r0>
(which also equates to saving power and money when said cycles are metered)
<f00b4r0>
robimarko: my builders see a x3 power usage increase when building
<f00b4r0>
that's the point.
<f00b4r0>
unless you suggests that x3 more electricity is irrelevant in the face of current market prices :)
<robimarko>
Well, that points again to need for some donations
<f00b4r0>
said donations will face the same constraints. Electricity is becoming a luxury ;P
<robimarko>
Well, they could help at least
<robimarko>
As the cost of the whole project is not really spread out
<f00b4r0>
not to mention that from an environmental PoV, we should strive to be more frugal maybe?
<robimarko>
Well, from environmental point, we should be using the most efficient HW
<ynezz>
and only on demand
<robimarko>
Yeah
<f00b4r0>
we do with what we have. My point is to do as good as we can with what we have
<ynezz>
f00b4r0: so what exactly do you suggest? Git pool for changes once a week?
<ynezz>
on stable/release branches?
<ynezz>
so we would just build check once a week?
<f00b4r0>
that'd be a start I guess
<f00b4r0>
I really don't see a need to build check release on every commit
<ynezz>
then it would depend on the number of changes in that time frame, how easy it would be to spot the source of potential build issue
<f00b4r0>
releases are nearly only backports after branching, aren't they? And they're branched from continuously-built master
<ynezz>
so having to inspect 2 changes or 30 makes difference, if we consider $500/hr price tag :)
<f00b4r0>
see above
<f00b4r0>
build test happens in master
<ynezz>
but the release branches are different, so it might work or might not
<f00b4r0>
they're also much lower commit rate
Borromini has quit [Ping timeout: 480 seconds]
<f00b4r0>
I mean, nearly every other foss project I can think of does nightly or weekly snapshots. Not sure why we would be so special that we can't do the same? :D
<ynezz>
can we back this with some real data and decide then? say after 3-6 months?
<ynezz>
quantify the number of builds, electricity/environment costs and see if it makes sense?
<ynezz>
f00b4r0: we barely do 48hrs nightly due to missing build workers/resources :)
<ynezz>
IIRC it takes more then 24hrs to build complete target family, so we pool the changes anyway
<f00b4r0>
sure. Looking at 22.03 in a 7-day window there has been ~9 commits (if I read the shortlog right and combine bulk commits into one). So that's potentially 8 builds saved by switching to weekly
<robimarko>
ynezz: Are all of the buildbots pretty much ran by OpenWrt or are some donations?
<ynezz>
workers are completely donations
<f00b4r0>
yes but what typically happens is that a build is started on commit A, then a few minutes/hours later commit B is added and the build already started on commit A must end before they pick up commit B. Those build are wasted and delay the newer ones.
<ynezz>
master is paid privately by nbd
<ynezz>
so dunno if that is counted as OpenWrt or donation :)
<f00b4r0>
ynezz: not sure why you say it cannot be "counted that way" tho?
<robimarko>
To me thats OpenWrt
<robimarko>
Donations would be 3rd party donated time or resources
<ynezz>
f00b4r0: what do you mean exactly by those 8 builds? complete target pool (72?) builds? Or 8 single target builds?
<robimarko>
zorun: Yeah, missed that fact
<f00b4r0>
ynezz: the former yes
<ynezz>
no, this is not the case
<f00b4r0>
(which gives an idea of how the problem "scales" :P)
<f00b4r0>
why not?
<zorun>
the thing is, for buildbot workers, cloud resources would be really really expensive
<zorun>
even with some free credits, it wouldn't work, we would burn through them quickly
<f00b4r0>
ynezz: not all 72 builds will happen sure
<f00b4r0>
but a fraction will
<ynezz>
f00b4r0: we don't even have resources to accomplish that, we would need 72 workers
<f00b4r0>
the fraction that starts in the interval between both commits
<ynezz>
yes, which you would need to closely monitor in enough long timeframe in order to make objective decision
<f00b4r0>
ok let me rephrase then
<f00b4r0>
assuming commits happen frequently enough to keep the workers 100% busy
<f00b4r0>
and assuming it takes 48h (worst case scenario) to build all targets, switching to weekly would essentially reduce the worker load by 5/7
<f00b4r0>
or am I missing something?
maxmadzz has joined #openwrt-devel
<f00b4r0>
imho the math is pretty simple, but maybe I'm mistaken ;)
indy has joined #openwrt-devel
<Tapper>
Is there ways that building OpenWrt can be sped up? Wen I run a build on my home PC my CPU spends a long time not doing mutch.
<PaulFertser>
Having everything in RAM helps.
<ynezz>
you can use prebuild external toolchain
borek has joined #openwrt-devel
borek has quit [Remote host closed the connection]
borek has joined #openwrt-devel
<Tapper>
II have 16 gig of ram. I know that is not mutch compared to a server.
<PaulFertser>
I've measured how much a minimal single target build from scratch takes on a dual-Xeon (Skylake SP) server with enough RAM. It was about 8 minutes, without external toolchain.
<Tapper>
I don't know how to use a prebuild external toolchain or how to build in ram.
<Tapper>
Any sorry I am getting OT.
<f00b4r0>
ynezz: i'm thinking another way to go at this without setting a fixed time interval would be to have a separate repo for the builders, where we would push changes only when we want them to be built. This way this can be both a cron job (weekly or otherwise) and a manual thing when a maintainer deems a full build necessary. And it shouldn't be overly difficult to implement either with minimal changes.
borek has quit [Remote host closed the connection]
<ynezz>
AFAIK building of toolchain is not paralel enough, so not utilizing maximum of the resources, once the toolchain is built, then it can utilize all 96 cores with 100% load
<f00b4r0>
PaulFertser: did you measure how much RAM that needed?
<ynezz>
this was done in RAM on GCE machine with 96 vCPUs
<ynezz>
same as with the disk space, so around 12-18GB, depends on platform
<PaulFertser>
Tapper: fwiw, for that test I booted the server with OpenWrt over PXE, then debootstrapped Debian on ramdisk, then cloned the repo and built.
<PaulFertser>
f00b4r0: unfortunately, no, but I can probably redo the test on similar hardware if needed.
<f00b4r0>
ynezz: same as diskspace + RAM usage for workload though
<f00b4r0>
https://imgur.com/a/mB6uSlk is the resource usage on slashdirt host when both of yesterday's builds where happening. I/O bottleneck is pretty clear
<ynezz>
"Imgur is temporarily over capacity. Please try again later."
<ynezz>
My notes from 2018: needs about 50GB of RAM as we're building in RAM, GCE n1-highcpu-96 86GB RAM $2.3/hr build time 34min, from scratch, no ccache etc. Today, you can do the same build in 31 min. on hetzner CPX51 VPS (AMD-EPYC 2nd GEN 16 vCPUs, 32GB of RAM, SSD disk?) for €0.1047/hr
<f00b4r0>
34mn to build everything?
<ynezz>
1st stage, firmware images and sdk
<f00b4r0>
for all targets?
<ynezz>
2nd stage, packages are different topic
<ynezz>
1 target, ipq40xx
<f00b4r0>
oh
<f00b4r0>
that's rather underwhelming then
<ynezz>
but you can spinup 72 workers for €7.2 and have complete release in one hour
<ynezz>
"So the complete rebuild of master and one stable release takes approximately 6 hours with 70 buildbot workers and costs 17.37€."
Tapper has quit [Ping timeout: 480 seconds]
goliath has joined #openwrt-devel
<ynezz>
I would say, that there is still like 20-40% room for optimizations, due to some manual fiddling etc.
<f00b4r0>
getting rid of rsync --checksum on 2nd stage would probably save some good time. This is extremely cpu and I/O intensive on both worker and master
<f00b4r0>
anyhow, I was just floating the idea of reducing our resource footprint, dunno if there's any interest. I'll submit a couple cleanups to phase2 master.cfg meanwhile :)
indy has quit [Ping timeout: 480 seconds]
indy_ has joined #openwrt-devel
<stintel>
lol ordered some WiFi 6E cards last year in October
<\x>
seems nice, and yeah man, oversized m.2 a thing now huh
<PaulFertser>
f00b4r0: is there some easy straightforward way I can log CPU and memory consumption during the build? Current master ath79 core image builds in 11m17.842s here.
<stintel>
\x: it's not m.2
<stintel>
or is it
<\x>
the one on the middle is
<stintel>
ah they have both
<\x>
A+E key
<\x>
also thats nice that they have external 5V pins on the oversized mini pcie ones
<stintel>
lol you scared me for a minute
<\x>
should be easy to use on x86, can always steal 5V from USB
<\x>
I cant understand this one, they say 1x1 and MU-MIMO
<\x>
how?
maxmadzz has quit [Ping timeout: 480 seconds]
guidosarducci_ has quit [Remote host closed the connection]
guidosarducci has joined #openwrt-devel
schwicht has joined #openwrt-devel
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
Misanthropos has joined #openwrt-devel
<f00b4r0>
PaulFertser: Probably creating a separate cgroup, with no particular restriction, and running the build within that cgroup. You then get all the accounting info in the various cgroup stat files
<f00b4r0>
have to run, bbl
<PaulFertser>
See you
<f00b4r0>
you'll probably want to test with the buildbot defconfig though (available as config.buildinfo on d.o.o), for things to be comparable
<f00b4r0>
really off now
dangole has joined #openwrt-devel
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
Borromini has joined #openwrt-devel
maxmadzz has joined #openwrt-devel
valku has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
dangole has joined #openwrt-devel
minimal has joined #openwrt-devel
<f00b4r0>
is there a quick way to duplicate a config section with uci?
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
* f00b4r0
notes that the 'interface' misnomer also plagues config/dhcp
aiyion has joined #openwrt-devel
aiyion has quit [Remote host closed the connection]
aiyion has joined #openwrt-devel
<soxrok2212>
i may have a few spare cpu cycles if needed?
<robimarko>
\x: Its 8devices, they just copy/paste stuff from the datasheet
<robimarko>
Sometimes completely wrong
Borromin1 has joined #openwrt-devel
<robimarko>
Sounds better if there MU-MIMO, though not really with a single chain
tlj_ has quit [Ping timeout: 480 seconds]
Borromini has quit [Ping timeout: 480 seconds]
Ansuel has joined #openwrt-devel
<stintel>
robimarko: do you know of an EU-based alternative to 8devices ?
<robimarko>
stintel: None unfortunately
<stintel>
guess I'll have to suck it up and deal with customs then
<Ansuel>
hi
<stintel>
o/
<robimarko>
stintel: Compex imports a lot of stuff
<robimarko>
Uggh, not Compex
<Ansuel>
day 3 (or 4?) of me against CONFIG_AUTOREMOVE
<robimarko>
Codico
<f00b4r0>
Ansuel: who's winning? ;)
<Ansuel>
the buildroot :(
<robimarko>
Ansuel: Be glad you dont deal with Gentoo, and trying to package that for embedded
<robimarko>
I envy those that just use Buildroot for products
<PaulFertser>
robimarko: how about OE?
<karlp>
stintel: lithuania _is_ in the EU, last I checked?
<robimarko>
PaulFertser: Havent really tried using it
<robimarko>
karlp: I think he meant an alternative that also in EU
<stintel>
exactly
<stintel>
the reason I tried 8devices is because it avoids customs crap
tlj_ has joined #openwrt-devel
<PaulFertser>
f00b4r0: some build log on a bare-metal server, unfortunately with pretty weak CPUs (140 W TDP) this time, I can do more tests if desired. https://paste.debian.net/hidden/2e8aa67c/
<PaulFertser>
(not sure how to get better statistics from cgroup)
Tapper has joined #openwrt-devel
tlj_ has quit [Remote host closed the connection]
dangole has quit [Ping timeout: 480 seconds]
<mrkiko>
stintel: what do you mean by "customs crap" ? BTW, english is not my language.
<mrkiko>
*first* language
<Ansuel>
tax
<Ansuel>
for import in eu
<mrkiko>
Ansuel: thanks, very clear
<mrkiko>
Ansuel: may I pm you?
<stintel>
the tax is not the problem
<Ansuel>
sure ?
<stintel>
dealing with Bulgarian customs is :)
<Ansuel>
stintel how much are they ?
<stintel>
EORI number and paperwork required
<stintel>
Ansuel: how much are what?
<Ansuel>
wow even paperwork ?
<stintel>
yes
<Ansuel>
customs
<Ansuel>
to handle all the paper
<stintel>
I fail to parse that question
<robimarko>
Hehe, in Croatia they delegated customs for regular post to Croatian Post
<robimarko>
Who now just guess at the value
<robimarko>
So I paid less now for expensive stuff than ever, its faster as well
<stintel>
I usually have stuff delivered to my parents in Belgium, I just get an SMS with a link to pay online, done
<stintel>
well, depends on the courier
<robimarko>
Thats way too easy
<stintel>
if it's bpost ... they lose your shipment
<robimarko>
You gotta pay in cash, thats the way in 2022
<Ansuel>
the future is nao
<robimarko>
And if its carriers, well good luck then
<robimarko>
They charge for everything and do nothing
<robimarko>
So its like 30 EUR for "their service"
<stintel>
but if I want it invoiced on my Bulgarian company it's another story
<stintel>
I should talk to my friend again, he runs a transport company here and can handle customs for me, if I give him power of attorney to do that
<stintel>
gladly pay him a small fee to never have to worry about that
<robimarko>
That is agood idea
<stintel>
also helped hurricos and me to get some SNIC10e to me without too much hassle :)
<stintel>
and then I replaced the broken hdd in his wife's laptop with an ssd and he took the shipping fees on him in return
valku1 has joined #openwrt-devel
valku has quit [Read error: Connection reset by peer]
valku1 is now known as valku
<Ansuel>
ok time to resume the fight against the buildroot.... this time with scripts and all automated
<f00b4r0>
PaulFertser: this looks like cgroup v2 structure which I'm not familiar with, but according to doc the memory high water mark is "memory.peak"
<Borromin1>
stintel: that much of a hassle to get electronics into Bulgaria?
Borromin1 is now known as Borromini
<stintel>
if customs are involved, it's more effort than I care to put in
<PaulFertser>
f00b4r0: memory.peak isn't present there
<f00b4r0>
ah right I meant to test that PR. Why don't days have 36h ;P
MaxSoniX has quit [Quit: Konversation terminated!]
<f00b4r0>
hmm, it doesn't apply cleanly
<Ansuel>
erm the qeen is dead... (sorry for ot) all the mems rip
<Borromini>
and in other news the ECB upped the interest rate ;)
<Ansuel>
speculation go brrrr
borek has quit [Ping timeout: 480 seconds]
minimal has quit [Quit: Leaving]
Tapper has quit [Ping timeout: 480 seconds]
dangole has joined #openwrt-devel
torv has quit [Remote host closed the connection]
torv has joined #openwrt-devel
<f00b4r0>
hmm
<f00b4r0>
qcom-ipq4018-wap-ac.dts:198.1-7 Label or path gmac0 not found
<f00b4r0>
I guess I'll have to ping robimarko
<Ansuel>
can you link the dts?
<f00b4r0>
i don't even know why this dts is built considering it's not the device I'm building for
<f00b4r0>
Ansuel: this is from #4721
<Ansuel>
what target is this?
<f00b4r0>
ipq40xx + DSA
<Ansuel>
oh well every dts is built
<f00b4r0>
wouldn't that be a problem if/when devices are marked "broken"?
<Ansuel>
broken doesn't mean have invalid dts
<Ansuel>
but yes it's a problem
cbeznea has quit [Quit: Leaving.]
clayface_ has joined #openwrt-devel
Misanthropos has quit [Ping timeout: 480 seconds]
Tapper has joined #openwrt-devel
noltari has joined #openwrt-devel
noltari_ has joined #openwrt-devel
Borromini has quit [Quit: Lost terminal]
noltari- has joined #openwrt-devel
noltari has quit [Ping timeout: 480 seconds]
noltari_ has quit [Ping timeout: 480 seconds]
Slimey_ has joined #openwrt-devel
Slimey_ is now known as Slimey
schwicht has joined #openwrt-devel
Misanthropos has joined #openwrt-devel
schwicht___ has quit [Ping timeout: 480 seconds]
bluew has joined #openwrt-devel
maxmadzz has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
schwicht_ has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
schwicht has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
schwicht_ has quit [Read error: No route to host]
maxmadzz has joined #openwrt-devel
bluew has quit [Quit: Leaving]
bluew has joined #openwrt-devel
noltari- has quit [Read error: Connection reset by peer]
noltari has joined #openwrt-devel
maxmadzz has quit [Ping timeout: 480 seconds]
Tapper has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]
maxmadzz has joined #openwrt-devel
russell-- has joined #openwrt-devel
minimal has joined #openwrt-devel
maxmadzz has quit [Ping timeout: 480 seconds]
Ansuel_ has joined #openwrt-devel
srslypascal is now known as Guest2312
srslypascal has joined #openwrt-devel
noltari has quit [Read error: Connection reset by peer]
Misanthropos has quit [Remote host closed the connection]
noltari has joined #openwrt-devel
<f00b4r0>
ynezz: FYI slashdirt bots failed with "connection lost" although they remain perfectly up and reachable. Not sure what happened.
<f00b4r0>
this apparently happened to azure ones too but these came back online, not the slashdirt ones
<f00b4r0>
containers are still running though
maxmadzz has joined #openwrt-devel
Guest2312 has quit [Ping timeout: 480 seconds]
Ansuel has quit [Ping timeout: 480 seconds]
Luke-Jr has joined #openwrt-devel
<f00b4r0>
oh forget that. Fiber down, link on backup ADSL. This explains that. *sigh*
maxmadzz has quit [Ping timeout: 480 seconds]
borek has joined #openwrt-devel
noltari_ has joined #openwrt-devel
Misanthropos has joined #openwrt-devel
borek has quit [Quit: borek]
borek has joined #openwrt-devel
borek has quit []
noltari has quit [Ping timeout: 480 seconds]
maxmadzz has joined #openwrt-devel
maxmadzz has quit [Remote host closed the connection]
<hurricos>
I have this mindnuming issue with avahi-daemon under Fedora, and I'm wondering if there's an mDNS advertisement service that OpenWrt supplies
<hurricos>
I know umdns exists, is there a way to dbus-wire-protocol bolt it onto something else?
<hurricos>
avahi-daemon is only broadcasting a requested record ONCE
<hurricos>
avahi-daemon is only broadcasting a requested record ONCE
<hurricos>
oops
<hurricos>
I'm just gonna collect umdns output, write some perl to convert it to an avahi service file, and call it a day
maxmadzz has joined #openwrt-devel
csrf has joined #openwrt-devel
kenny has joined #openwrt-devel
rsalvaterra has quit []
rsalvaterra has joined #openwrt-devel
maxmadzz has quit [Read error: Connection reset by peer]