Werner changed the topic of #armbian to: armbian - Linux for ARM development boards | www.armbian.com | Github: github.com/armbian | Commits: #armbian-commits | Developer talk: #armbian-devel | Forum/Twitter feed: #armbian-rss | Logs: -> irc.armbian.com
<schwarz_Kat> lanefu: on IRC earlier when I asked about wanting to try to restore the GPIO sysfs functionality to the orangepizeroplus.
<schwarz_Kat> Build shows [ o.k. ] Using kernel config file [ config/kernel/linux-sunxi64-current.config ]
<schwarz_Kat> Unless I'm mistaken (newbie) the config used by orangepizeroplus is linux-sunxi64-current.config.
<schwarz_Kat> At first I thought that the issue could be missing CONFIG_GPIO_SYSFS=y, but building with this didn't fix the issue.
<ArmbianHelper> ^ Error retrieving title. Check the log for more details.
<schwarz_Kat> Then... **I found a build that I did a couple of months back that was kernel 5.15.25 and had a boot config-5.15.25-sunxi64 that matched the current build**.
<schwarz_Kat> So I don't think that the issue is the kernel config. I tried to disect the differences between the two builds, but the one that I'm comparing with has had a whole lot of extra stuff installed, making it difficult. I couldn't find any significant differences in modules. Device tree on boot matches.
<schwarz_Kat> I then tried to create a branch with head on an earlier commit. I wasn't able to go back far enough, though I've only recently started using git, so I'm still pretty clueless.
<schwarz_Kat> I then thought that I could try and look at differences in patches between legacy and current... not even vaguely similar seems to have been completely reworked.
<schwarz_Kat> Anyone able to offer some hints at what I should try next?
califax has quit [Remote host closed the connection]
califax has joined #armbian
<Armbian-Discord> <l​anefu> Is there an device tree overlay that needs to be enable via armbian-config to activate gpio?
<Armbian-Discord> <R​ichN> what are you enabling 1 wire
<Armbian-Discord> <R​ichN> you have to use the armbian-config > system > Hardware
<Armbian-Discord> <R​ichN> and enable the 1 wire and maybe i2c
norwich_ has joined #armbian
norwich has quit [Ping timeout: 240 seconds]
norwich_ is now known as norwich
<Armbian-Discord> <l​anefu> They just want standard gpio over sysfs
<schwarz_Kat> lanefu: I don't see a dto that would do that. In the 5.15.25 board that has the sysfs filesystem, there are no overlays set for that in armbianEnv.txt
TRS-80 has quit [Quit: WeeChat 3.0]
shoragan has quit [Ping timeout: 240 seconds]
<schwarz_Kat> RichN: No not 1 wire. GPIO sysfs. Deprecated in the kernel, but still very much in use, and until now available on this board. I suspect that it dropped off with the latest version changes. It's available on a build that I did on what was then the edge build just before the change. Now it is only available on the legacy build. I'm trying to understand what has dropped off.
<lanefu> So do you see the gpio flcjip in sysfs
<lanefu> You do have to enable / export the pins manually they won't show up on their own
<schwarz_Kat> lanefu: the main symptom is no /sys/class/gpio at all so no enable / export.
<Armbian-Discord> <l​anefu> Bummer. Yeah sorry I'm not closer to stuff to look around
<Armbian-Discord> <l​anefu> armbian monitor -m
<schwarz_Kat> I will have to look at that tomorrow.
<schwarz_Kat> Also going to try to find the device(s) and drivers on a working system. WML.
xoan8 has joined #armbian
xoan has quit [Ping timeout: 272 seconds]
xoan8 is now known as xoan
samythemany1 has joined #armbian
samythemany11 has joined #armbian
samythemany has quit [Ping timeout: 240 seconds]
samythemany1 has quit [Ping timeout: 240 seconds]
adeepv has joined #armbian
schwarz_Kat has left #armbian [Offline]
xoan5 has joined #armbian
<Armbian-Discord> <Q​ueen Fiona> Hey, am I correct in hearing the Orange Pi PC is no longer CSC?
<Armbian-Discord> <Q​ueen Fiona> Also, what's new in the 5.17 kernel where Armbian boards are concerned?
xoan has quit [Ping timeout: 260 seconds]
xoan5 is now known as xoan
archetech has joined #armbian
smcavoy has quit [Read error: Connection reset by peer]
smcavoy3 has joined #armbian
Mony has quit [Quit: ZNC - http://znc.in]
Mony has joined #armbian
Mony has quit [Client Quit]
Mony has joined #armbian
kmcopper has joined #armbian
Smedles has quit [Quit: http://quassel-irc.org - Chat comfortably. Anywhere.]
Smedles has joined #armbian
<ArmbianHelper> ^ Bugfix - wrongly linked folders when adding CLI PPA example by igorpecovnik · Pull Request #3622 · armbian/build · GitHub
aasami_ has joined #armbian
<aasami_> How are partitions mounted at start? Automatically. Mentioning fstab is useless.
<aasami_> With startupscript is not ar all wanted
<aasami_> ar = at
<aasami_> By kernel it is not wanted too because it doesn't work
<IgorPec> which partitions?
kmcopper has quit [Remote host closed the connection]
samythemany11 has quit [Quit: Bridge terminating on SIGTERM]
samythemany has joined #armbian
alekksander has joined #armbian
alekksander has quit [Remote host closed the connection]
alekksander has joined #armbian
samythemany has quit [Quit: Bridge terminating on SIGTERM]
samythemany has joined #armbian
wd has joined #armbian
<aasami_> IgorPec: sda1, sda2, sda5, sda7, sda8, sdb5, sdb5, sdc5, sdc6, sdc7, sdd1, sdd2
<aasami_> IgorPec: UUID=cb4d93b6-32b7-47e7-8fa7-a2a64ea34ad5 none swap sw 0 0
<aasami_> IgorPec: UUID=7295b79d-3fb4-4e6e-a62e-026a6950ced3 / ext4 errors=remount-ro 0 1 works fine :)
<aasami_> IgorPec: UUID=42fbd3c9-e846-4d12-8962-8cc28c13d213 /mnt/sda5 ext4 0 1 never works, stops booting
<FergusL> C2 with unreliable sd slot update: boots fine with an emmc module
<FergusL> Only tried the odroid 16.04 image it came with so far. Flashing armbian for testing now
<Herc> FergusL, I never had any SD issues with my two odroidC2
<Herc> emmc was faster ofc, but I lost them somehow, lol
<aasami_> IgorPec: mounting manually works fine: mount -t ext4 /dev/sda5 /mnt/sda5
<FergusL> Herc lost the emmcs ?!!
<FergusL> I agree it's weird though, hence why I decided to dig the issue
<FergusL> but now I desperately want an rk3399 board :(
<Armbian-Discord> <I​gorPec> aasami_ automouting probably failed when devices are not initialized fast enough. google for solution.
<ArmbianHelper> ^ Fixing deboostrap fail due to too old ca-certificate package by igorpecovnik · Pull Request #3624 · armbian/build · GitHub
<IgorPec> need review
archetech has quit [Quit: Konversation terminated!]
<Armbian-Discord> <n​iccodemi> I am trying to compile image for Bananpi M1 board which has broken sd card detect pin. I added line CONFIG_MMC0_CD_PIN="" to the end of "Bananapi_defconfig" and created patch. When building image I get following warning "[\e[33mu\e0m u-boot-sunxi-current.patch failed wrn". Any idea what is the issue?
archetech_n2 has joined #armbian
SallyAhaj has quit [Remote host closed the connection]
<Armbian-Discord> <r​pardini> IgorPec: taking a look at AR-1140 (ca-certificate stuff on focal)
<ArmbianHelper> AR-1140 [Bug] "Deboostrap fails due to too old ca-certificate package" reported by Igor Pecovnik at 2022-04-02. Status: In Progress
<Armbian-Discord> <r​pardini> it is a bit wierd, considering ca-certificates is not in restricted...
<IgorPec> ha, strange. but this fix helps
<Armbian-Discord> <r​pardini> version in focal and focal-updates is actually the same
<IgorPec> let me do one more test
<Armbian-Discord> <r​pardini> but, here's the thing https://www.ssllabs.com/ssltest/analyze.html?d=imola.armbian.com&hideResults=on
<ArmbianHelper> ^ SSL Server Test: imola.armbian.com (Powered by Qualys SSL Labs)
<Armbian-Discord> <r​pardini> imola has indeed invalid certificates installed on one of the certification paths
<Armbian-Discord> <r​pardini> so might be easier/better just to fix imola's cert
<IgorPec> ahaa
<ArmbianHelper> ^ DST Root CA X3 Expiration (September 2021) - Let's Encrypt
<Armbian-Discord> <r​pardini> my guess is error is random, 2 cert paths, one fails, adding restricted or not probably does not make a difference, just good/bad luck during the tries... 🤔
<IgorPec> hm, tl;dr; what is the problem?
<IgorPec> openssl version?
<Armbian-Discord> <r​pardini> dunno how/what is serving imola's 443 port... but I guess it is using old cacert or intermediaries
<Armbian-Discord> <r​pardini> cert itself is correct, but intermediaries not
<Armbian-Discord> <r​pardini> someone will for sure correct me
<IgorPec> this is vm machine here at my server, i have setup this www some time ago
<IgorPec> its pure ubuntu 20.04 and that python for auto cert renew
<Armbian-Discord> <r​pardini> certbot maybe?
<IgorPec> yes
<IgorPec> i think
<IgorPec> there are two subdomains
<IgorPec> what would be causing this. don't understand quite
<Armbian-Discord> <r​pardini> I am investigating a bit more
<Armbian-Discord> <r​pardini> comparing to some LE setups I have
<Armbian-Discord> <r​pardini> actually my setups are exactly the same 😐 so not a server problem indeed
<Armbian-Discord> <r​pardini> might be openssl version related
<Armbian-Discord> <r​pardini> (on the client/debootstrap)
<IgorPec> then perhaps this is taken from restricted?
<Armbian-Discord> <r​pardini> still does not explain why adding restricted solves it though 😉
<Armbian-Discord> <r​pardini> lemme check
<IgorPec> i have 10/10 runs with added restricted
<IgorPec> always works, without always fail ;)
<Armbian-Discord> <r​pardini> yeah, gotta be libssl or something related
<Armbian-Discord> <r​pardini> what led you to try to add restricted in the first place? not a random try I guess
<IgorPec> i saw after deebstrap finishes packages are getting updated
<IgorPec> and this was one of the tries, more like why this is not there in first place
<IgorPec> restricted repo we anyway have
<Armbian-Discord> <r​pardini> hmm, indeed, we add restricted afterwards anyway
<IgorPec> and i ran into this problem on the machine
<IgorPec> if you install some older image, you can't update from armbian repo without updatoing ca-certificates first
<IgorPec> if this tells you anything
<Armbian-Discord> <r​pardini> approved PR, "why" is a bit over my head but makes sense.
<Armbian-Discord> <r​pardini> probably has to do with ca-certificates (list of certs) vs openssl (combining list of certs into big bundle during install)
<IgorPec> probably.
<Armbian-Discord> <I​gorPec> btw, i have fixed that repository making in with github pages
<Armbian-Discord> <I​gorPec> wrong key naming ... lame 😉
<Armbian-Discord> <r​pardini> ahn, I had forgot about that with vacations (good sign)
<Armbian-Discord> <r​pardini> what was the problem in the end? I couldn't for the life of me find it
<Armbian-Discord> <I​gorPec> just wrong variable / secret name
<Armbian-Discord> <r​pardini> damn.
<Armbian-Discord> <I​gorPec> works now, update repo on code change. now in monday i have a meeting to start moving that armbian-config
<Armbian-Discord> <r​pardini> ahn ok!! found it, wrong name when passing $secret to crazymax's thingy
<Armbian-Discord> <r​pardini> damned that flew very low on my radar
<Armbian-Discord> <I​gorPec> this restricted hack works for me, but not in CI
<Armbian-Discord> <r​pardini> lemme build a focal see if I can reproduce
<Armbian-Discord> <I​gorPec> one possible relation is our redirector
<Armbian-Discord> <I​gorPec> there are some changes
<Armbian-Discord> <r​pardini> yeah, unlikely but could be, if redirecting randomly to something with a broken cert
<Armbian-Discord> <I​gorPec> testing now
<Armbian-Discord> <r​pardini> those PPA's, they come from BUILD_DESKTOP=yes I guess?
<Armbian-Discord> <r​pardini> btw great stuff with nvidia and chrome for amd64 desktops 👍
<Armbian-Discord> <I​gorPec> yes, also probably would be nice to add amd drivers
<Armbian-Discord> <r​pardini> we've already the open source ones and firmware in the -full version
<Armbian-Discord> <r​pardini> for amdgpu
<Armbian-Discord> <I​gorPec> build desktop yes are some, just one is on cli but just jammy / bullseye
<Armbian-Discord> <I​gorPec> aha ok
<Armbian-Discord> <r​pardini> I did for the Phythium a few months ago, don't remember if I pushed -full firmware though
<Armbian-Discord> <r​pardini> did yes.
<Armbian-Discord> <r​pardini> I've a radeon and works perfect out of the box first-boot
<Armbian-Discord> <r​pardini> nvidia used to require the dkms, but now you fixed that too 😉
<Armbian-Discord> <I​gorPec> i only have one nvidia here in the workstation
<Armbian-Discord> <I​gorPec> have to move to armbian one day 😉
<Armbian-Discord> <r​pardini> amd64 really helps with with userland development even for arm64
<Armbian-Discord> <r​pardini> I come up with a scheme useing qemu and pci passthrough
alekksander has quit [Ping timeout: 246 seconds]
<Armbian-Discord> <r​pardini> build amd64 image, boot it in qemu passing trhough an USB controller via pci passthrough
<Armbian-Discord> <I​gorPec> for gaming purposes ?
<stipa> for userland dev
<Armbian-Discord> <r​pardini> ahn for gaming I do too, pci-e passthrough of the radeon, but guest is Windows
<Armbian-Discord> <r​pardini> (host is Armbian amd64 though so it counts)
<Armbian-Discord> <I​gorPec> yes that
<stipa> what about those 3K ARM workstations, is that going to be supported ?
<stipa> 3K $
<IgorPec> it probably just works, but no fine tunning
<stipa> i see they always come and go
<IgorPec> nobody send us any
<stipa> may they expect that you'll buy some
<Armbian-Discord> <r​pardini> I've hearsay reports of it working on 128-core Amperes
<Armbian-Discord> <r​pardini> bare-metal
<Armbian-Discord> <r​pardini> I myself have no access to any yet
<IgorPec> lanefu played something with that afaik
<IgorPec> i think we have some arm server somewhere
<IgorPec> but not 128c, something less
<stipa> maybe 32
<Armbian-Discord> <r​pardini> someone reached out a time ago with good results on Altra Max, but when I asked for evidence/neofetch they disappeared
<stipa> heh
<Armbian-Discord> <r​pardini> really if the boards uses decent UEFI (Tianocore / edk2) as most does it should just work.
<Armbian-Discord> <r​pardini> only board I've heard of problems with unfortunately was the SolidRun one
<stipa> it's hard to tell without an access to a server like that, maybe it's mainline
<Armbian-Discord> <r​pardini> but that is 16x A-76
<Armbian-Discord> <r​pardini> they promised to send me a board to investigate, but it seems they're really 100% out of anything to sell much less give away
<IgorPec> i heard they can't lead anything
<IgorPec> but that's 6m old info
<stipa> low interest probably
<IgorPec> a combinatoion probably
<Armbian-Discord> <r​pardini> 4x 10Gbit SFP ports very much interest me, and others, but they seem to have supply problems
<Armbian-Discord> <r​pardini> as most everyone else.
<stipa> yeah, but if the interest is so low i have no idea how they can even finance a prototype
<Armbian-Discord> <r​pardini> yeah. so right now my validation is still VMs. AWS Graviton2, OCI's Ampere all work. One day I might shell a few bucks to try on a bare-metal AWS Graviton2...
<Armbian-Discord> <r​pardini> stipa: ha, found the "3K" you mention, is rack version of SolidRun Honeycomb
<Armbian-Discord> <r​pardini> literally USD 3000
<Armbian-Discord> <r​pardini> so it's two of the honeycomb squashed together (2x 16-core A76)
<Armbian-Discord> <r​pardini> I guess problems would be exactly the same, requires patches to both kernel and userspace (libc!) to work
<stipa> rpardini , i meant on this one https://www.asacomputers.com/Cavium-ThunderX-ARM.html
<ArmbianHelper> ^ Marvell ThunderX | Professional Workstation | ASA Computers
<Armbian-Discord> <r​pardini> oh yes, that is a bit older, I've a few confirmations of it working
xoan1 has joined #armbian
<Armbian-Discord> <r​pardini> are not really super fast cores, I'd say 1/2 a Graviton2 core, or somewhat slower than Oracle's Ampere core
sunshavi has joined #armbian
<stipa> oh
<Armbian-Discord> <r​pardini> but yeah, 48 of those make for a nice build server.
<Armbian-Discord> <r​pardini> I did not follow what happened to Cavium, it seems it's Marvell now
xoan has quit [Ping timeout: 260 seconds]
xoan1 is now known as xoan
<stipa> could be
<Armbian-Discord> <r​pardini> well, off to rebase armbian-next onto master again... always a nice adventure
<IgorPec> ok, one explanation for failing cache creation
<IgorPec> this is created natively, not in container
<IgorPec> is is not under full control. well, perhaps i will just refactor this ... be back. must do some phy relocation
<Armbian-Discord> <r​pardini> hmm but error happens after deboostrap, when running apt under chroot
<Armbian-Discord> <r​pardini> I still did not reproduce it
Myron has joined #armbian
Myron has quit [Killed (NickServ (GHOST command used by guest6191!~Myron@94.6.204.7))]
Myron has joined #armbian
<Armbian-Discord> <I​gorPec> this fails for me always https://pastebin.com/ATy9Ppab
<ArmbianHelper> ^ ./compile.sh BOARD="bananapi" BRANCH="current" RELEASE="focal" BUILD_MINIMAL="no - Pastebin.com
<Armbian-Discord> <R​ichN> thete is a script that will make it easier to follow and build with ..
<IgorPec> Get:1 http://ports.ubuntu.com focal-security/main armhf ca-certificates all 20210119~20.04.2 [145 kB]
<ArmbianHelper> ^ Index of /
<IgorPec> @rpardini security repo has that newer version, which does work
<Armbian-Discord> <r​pardini> hmm security yes, makes sense... but what relation to restricted?
<IgorPec> none i guess
<Armbian-Discord> <r​pardini> maybe one implies the other
<IgorPec> adding security to deboostrap and see if this fixes it
<Armbian-Discord> <l​anefu> Yeah I don't think the honeycombs are a good buy brand new
<Armbian-Discord> <r​pardini> oh hi there @lanefu
<Armbian-Discord> <r​pardini> you're talking now to rpardini, CKA'd by the Linux Foundation
Some-body_ has joined #armbian
DarthGandalf has quit [Ping timeout: 268 seconds]
Some-body_ is now known as DarthGandalf
<Armbian-Discord> <r​pardini> this fails in so many different ways for me it's not even funny. but none with https error, yet. still investigating
<Armbian-Discord> <I​gorPec> yeah, i can't explain it yet neither
<Armbian-Discord> <r​pardini> it's gotta be the redirector Igor
<Armbian-Discord> <I​gorPec> hmm, i am still novice with its configuring
<Armbian-Discord> <r​pardini> simple reason, we use http:// for the apt source
<Armbian-Discord> <I​gorPec> @cats @lanefu
<Armbian-Discord> <I​gorPec> if they are around to inspect
<ArmbianHelper> ^ build/general.sh at master · armbian/build · GitHub
<Armbian-Discord> <r​pardini> so redirector is sending http -> https in this case
<Armbian-Discord> <I​gorPec> correct
<Armbian-Discord> <r​pardini> which I guess it shouldn't to begin with...?
<Armbian-Discord> <I​gorPec> did you try with hard coded values to fixed host?
<Armbian-Discord> <I​gorPec> to bypass redirector
<Armbian-Discord> <r​pardini> gonna hardcode imola there now
<Armbian-Discord> <r​pardini> bypassed, but worked. rebuilding rootfs cache now
<Armbian-Discord> <r​pardini> bingo reproduced
<Armbian-Discord> <r​pardini> Certificate verification failed: The certificate is NOT trusted. The certificate chain uses expired certificate. Could not handshake: Error in the certificate verification. [IP: ::1 3142]
<Armbian-Discord> <r​pardini> so it is only during rootfs cache rebuild
<Armbian-Discord> <I​gorPec> what is here different?
<Armbian-Discord> <r​pardini> nah, I'm an idiot.
<Armbian-Discord> <r​pardini> of course if I had previous working cache, then it would still work
<Armbian-Discord> <I​gorPec> yep 😉
<Armbian-Discord> <r​pardini> now investigating why this breaks now
<Armbian-Discord> <r​pardini> yeah, during debootstrap it's Retrieving ca-certificates 20190110ubuntu1
<Armbian-Discord> <r​pardini> which is ancient
<Armbian-Discord> <r​pardini> updated one is in security repo
<Armbian-Discord> <r​pardini> so bug was there for a decade
<Armbian-Discord> <r​pardini> redirector now sent http to https, and made it manifest
<Armbian-Discord> <I​gorPec> yeah, that is root cause, now what to do?
<Armbian-Discord> <r​pardini> we'd need to split this, do apt update+upgrade separately/before adding all the other sources
<Armbian-Discord> <r​pardini> so that security gets added and upgraded, ca-certificates fixed, then we add the other repos
<Armbian-Discord> <r​pardini> or: https://gitlab.mister-muffin.de/josch/mmdebstrap 😉
<ArmbianHelper> ^ josch/mmdebstrap - mmdebstrap - Muffin Gitea
Myron has left #armbian [#armbian]
<Armbian-Discord> <r​pardini> mmdebstrap promises to allow debootstrap to use multiple repos from the start, so saves time
<Armbian-Discord> <I​gorPec> i see, yes
<Armbian-Discord> <r​pardini> for now I'd split, or fix redirector 😉
<Armbian-Discord> <I​gorPec> perhapts this direction? do we have guts ?
<Armbian-Discord> <I​gorPec> wait, perhaps imola is the problem
<Armbian-Discord> <r​pardini> I've too much half-finished, would not responsible of me to engage in this right now, but would in a month or so 😉
<Armbian-Discord> <I​gorPec> yes, its auto redirecting to https
<Armbian-Discord> <r​pardini> nope, just checked. in that state (with old ca-certificate) none if my https servers work either
<Armbian-Discord> <r​pardini> so it's not imola
<Armbian-Discord> <I​gorPec> ok
<Armbian-Discord> <I​gorPec> then workaround
<Armbian-Discord> <r​pardini> still pondering how your addition of restricted somehow fixed it, even if only on your local machine
<Armbian-Discord> <r​pardini> cause fix is in security which is separate repo, not component of main repo
<Armbian-Discord> <I​gorPec> i wouldn't hold on that much. perhaps i was using BETA="no"
<Armbian-Discord> <I​gorPec> and that uses different redirector logic
<Armbian-Discord> <r​pardini> ahn, and then maybe in that case redirector is different
<Armbian-Discord> <I​gorPec> i think only dl and apt are using new one . not 100% sure
<Armbian-Discord> <I​gorPec> this is pretty much WIP
alekksander has joined #armbian
<Armbian-Discord> <r​pardini> btw if redirector send http -> https, it negates the cache of apt-cacher-ng
<Armbian-Discord> <r​pardini> so even if workaround I'd say redirector needs fix 😉 @cats
SallyAhaj has joined #armbian
DarthGandalf has quit [Quit: Bye]
DarthGandalf has joined #armbian
<Armbian-Discord> <c​ats> Is the redirector not working properly?
<Armbian-Discord> <c​ats> http redirection works fine from what I can see
<Armbian-Discord> <c​ats> got any examples?
<ArmbianHelper> (REDIRECT) ^ Index of /armbian/
<ArmbianHelper> ^ Error retrieving title. Check the log for more details.
<Armbian-Discord> <c​ats> if the build script is calling it over https it'll be https
<Armbian-Discord> <c​ats> if it's http it'll be http
<Armbian-Discord> <c​ats> any mirrors that use auto https redirection need to be fixed because that's a requirement to be a mirror imo
<Armbian-Discord> <r​pardini> seems to be redirecting http to https in some situations
<Armbian-Discord> <c​ats> If you have any examples I'd love to debug it :)
<Armbian-Discord> <c​ats> I can't find anything YET but I'll dig into the code
<Armbian-Discord> <r​pardini> by example, you mean...?
<Armbian-Discord> <c​ats> yup
<Armbian-Discord> <c​ats> like
<Armbian-Discord> <c​ats> a url that it redirects from http -> https
<Armbian-Discord> <c​ats> start and end url
<Armbian-Discord> <r​pardini> let me gather the facts for ya then
<Armbian-Discord> <c​ats> if you can put a debug call to print the url before it requests it that'd be awesome
<Armbian-Discord> <c​ats> If you have a command I can run I can also run it myself with some debugging in place
<Armbian-Discord> <r​pardini> maybe easier than that. take a look https://github.com/armbian/build/blob/master/lib/general.sh#L250
<ArmbianHelper> ^ build/general.sh at master · armbian/build · GitHub
<Armbian-Discord> <r​pardini> that is fixed to http://
<Armbian-Discord> <r​pardini> but we're running into TLS cert trouble...
<Armbian-Discord> <r​pardini> not always
<Armbian-Discord> <c​ats> I'm wondering if certain mirrors force https
<Armbian-Discord> <c​ats> I can detect that and kick them out of the pool
<Armbian-Discord> <r​pardini> so it's a bit hard to pinpoint, but it seems imola was one http to https response
<Armbian-Discord> <r​pardini> also: if http to https response, we lose apt-cacher-ng cache's effectiveness
<Armbian-Discord> <r​pardini> hmm. by force https you mean the mirrors themselves redirect to https, not the redirector? 🤔
<Armbian-Discord> <r​pardini> goes testing again
<Armbian-Discord> <c​ats> Yup, the mirrors themselves will
<Armbian-Discord> <c​ats> the redirector passes through the scheme
<Armbian-Discord> <r​pardini> that is not the case, at least with imola; that keeps whatever I send it (by hardcoding / skipping redirector)
<Armbian-Discord> <c​ats> hmmm
<Armbian-Discord> <r​pardini> I myself don't get redirected to imola... was happening to Igor
<Armbian-Discord> <r​pardini> but also in CI
<Armbian-Discord> <r​pardini> sorry for lack of evidence, should have collected a curl
<Armbian-Discord> <c​ats> it's fine, I'm looking into potential edge cases
<Armbian-Discord> <c​ats> the new redirector was meant to fix the inverse problem lol
<Armbian-Discord> <r​pardini> yeah, that's what I thought too
<Armbian-Discord> <c​ats> is it only appearing when it's used by apt?
<Armbian-Discord> <r​pardini> yeah. but the symptom is a very far away, a TLS cert error due to ancient ca-certificates
<Armbian-Discord> <r​pardini> only on ancient focal and right after initial debootstrap, before first apt update
<Armbian-Discord> <c​ats> I'm running the build script (if that'll reproduce it?) and I'll see if I can get the output of the packages
<Armbian-Discord> <r​pardini> so quite intricate, but the point is: it was supposed to be http and thus be a non-problem
<Armbian-Discord> <c​ats> actually, would you be comfortable sharing your IP address in DM? I could probably check the logs to see which urls it hit and try to replicate it
<Armbian-Discord> <r​pardini> yeah but mine works!
<Armbian-Discord> <r​pardini> it's Igor's 😉 he's "physically moving things around" I think
<Armbian-Discord> <c​ats> ahhh
<Armbian-Discord> <r​pardini> I'm gonna it now
<Armbian-Discord> <r​pardini> Ahn!
<Armbian-Discord> <r​pardini> it might be BETA or not too!
<Armbian-Discord> <r​pardini> (Igor uses that stuff, I don't, normally, unless trying to reproduce Igor)
<Armbian-Discord> <c​ats> Igor had the pastebin earlier which is confirmed not working, perfect
<Armbian-Discord> <c​ats> lol, VM I'm trying to build it on ran out of space, oops
<Armbian-Discord> <r​pardini> I just did a run with beta that worked too Get:1 http://beta.armbian.com focal/main amd64 linux-image-current-x86 amd64 22.05.0-trunk.0020 [69.2 MB]
<ArmbianHelper> ^ armbian storage
<Armbian-Discord> <c​ats> beta is still using the old redirector
<Armbian-Discord> <c​ats> perhaps next time we'll start with beta and archive lol
<Armbian-Discord> <r​pardini> I'm trying both...
<Armbian-Discord> <r​pardini> that desktop build is a monster though
<Armbian-Discord> <c​ats> yeah, I gave my VM 64GB more space through some dirty fdisk editing lol
<Armbian-Discord> <c​ats> I didn't feel like mounting a live cd to resize it
<Armbian-Discord> <r​pardini> eh. I've gotta one day deliver my ready-to-build VM image
<Armbian-Discord> <r​pardini> I made it work once, inception mode and all
<Armbian-Discord> <c​ats> I haven't run into any issues yet
<Armbian-Discord> <c​ats> just unpacked all the packages
<ArmbianHelper> ^ [image/png] (643.2KiB)
<Armbian-Discord> <r​pardini> yeah apt.armbian.com works for me too. I'm stumped
<Armbian-Discord> <r​pardini> did you get at https://armbian.atlassian.net/browse/AR-1140 ?
<ArmbianHelper> AR-1140 [Bug] "Deboostrap fails due to too old ca-certificate package" reported by Igor Pecovnik at 2022-04-02. Status: In Progress
<ArmbianHelper> ^ [AR-1140] - Jira
<Armbian-Discord> <c​ats> Yeah that looks like imola
<Armbian-Discord> <c​ats> 1381Get:38 https://cli.github.com/packages stable/main amd64 Packages [338 B]
<ArmbianHelper> ^ Error retrieving title. Check the log for more details.
<Armbian-Discord> <c​ats> potentially that too
<Armbian-Discord> <r​pardini> nah, that's from a PPA, and that is https:// from the start
<Armbian-Discord> <r​pardini> it's mostly "how did imola get there" as https, since we add repo as http
<Armbian-Discord> <c​ats> yeah, igor said imola had an improper https redirect
<Armbian-Discord> <c​ats> (on the server config side)
<Armbian-Discord> <c​ats> So it was redirecting twice, once from apt.armbian.com -> imola, then imola saw http and went https
<Armbian-Discord> <r​pardini> well then it's disappeared. thanks a lot for taking a look @cats
<Armbian-Discord> <c​ats> sure thing, let me know if it appears again!
<Armbian-Discord> <c​ats> I was hoping the redirector would help the project and it seems like it has lol
<Armbian-Discord> <c​ats> albeit with some pain points getting it implemented
<Armbian-Discord> <c​ats> I need to do proper unit testing, I was lazy and didn't...
<Armbian-Discord> <c​ats> (it is open source btw, just not on github)
<Armbian-Discord> <c​ats> https://git.meow.tf/tyler/armbian-router
<ArmbianHelper> ^ tyler/armbian-router: A simple http router for apt downloads based on GeoIP - armbian-router - Gitea: Git with a cup of tea
alekksander has quit [Quit: Konversation terminated!]
<Armbian-Discord> <r​pardini> nice stuff, I didn't know any of this existed
<Armbian-Discord> <r​pardini> armbian is infinite fun in all areas
<Armbian-Discord> <c​ats> Yeah, this code is a bit more gross but it does geoip location redirection, not just country/continent
<Armbian-Discord> <c​ats> So it takes the top x servers, balances requests to them based on weighting
<Armbian-Discord> <c​ats> (so in theory we could prioritize better performing servers/servers with more capacity if there's a bottleneck at any point)
<Armbian-Discord> <c​ats> and also live monitoring of statuses, so we shouldn't have any dead mirrors sitting in there for more than 60 seconds
<Armbian-Discord> <r​pardini> nice. so this listens http, some nginx in front terminates tls and proxies to it, X-Forwarded stuff et al?
<Armbian-Discord> <c​ats> Yup
<Armbian-Discord> <I​gorPec> tl; dr; my workaround still needed?
<Armbian-Discord> <c​ats> the imola fix yes
<Armbian-Discord> <c​ats> tl;dr is that we need to make sure no mirrors use forced https redirects, but I've added monitoring for that on the redirector to kick those out of the pool
<Armbian-Discord> <I​gorPec> i can disable forced https redirect
<Armbian-Discord> <c​ats> yup, mirrors will need to have it disabled sadly
<Armbian-Discord> <c​ats> while forced https is good for security, it breaks http downloads for mirrors to handle that
<Armbian-Discord> <r​pardini> yeah, https negates the local/net cache we use, which is a proxy
<Armbian-Discord> <c​ats> if you're forcing your build script to use a mirror, make sure it's http:// and not https:// as well
<Armbian-Discord> <r​pardini> yeah http:// is forced there since forever, it's hardcoded
<Armbian-Discord> <r​pardini> can't change it
<Armbian-Discord> <c​ats> if it's a different mirror than default, that is
<Armbian-Discord> <r​pardini> it's required for the proxy to be effective
<Armbian-Discord> <c​ats> idk if Igor has a local redirect to force build scripts to use his local setup
<Armbian-Discord> <r​pardini> (we've apt-cacher-ng proxy for builds)
<Armbian-Discord> <r​pardini> so in the end if https is used, cache is skipped and BW usage skyrockets
<Armbian-Discord> <c​ats> makes sense
<Armbian-Discord> <c​ats> when you're building 50 images it adds up lol
<Armbian-Discord> <r​pardini> I'm building 50 images per hour or more when working on this
<Armbian-Discord> <c​ats> dang
<Armbian-Discord> <r​pardini> a CI run must build, dunno, 200 images
<Armbian-Discord> <r​pardini> Igor will know
<Armbian-Discord> <I​gorPec> yes, i do have
<Armbian-Discord> <I​gorPec> forcing local mirror to fasten things
<Armbian-Discord> <r​pardini> anyway it was Igor's own fault who called you here, ran away, fixed it sneakly and now is back laughing at me
<Armbian-Discord> <c​ats> hahahaha
<Armbian-Discord> <r​pardini> 😄
<Armbian-Discord> <r​pardini> was fun we had a nice chat/debug session so thanks! 😄
<Armbian-Discord> <I​gorPec> yeah
<Armbian-Discord> <c​ats> :twoboldSip:
<Armbian-Discord> <l​anefu> LOL
<Armbian-Discord> <I​gorPec> i have disabled https redirection on imola
<Armbian-Discord> <I​gorPec> You should be redirected automatically to target URL: <a href="https://imola.armbian.com/beta/
<ArmbianHelper> ^ armbian storage
<Armbian-Discord> <I​gorPec> should this redirect to http
<Armbian-Discord> <I​gorPec> this comes from curl http://beta.armbian.com
<ArmbianHelper> ^ armbian storage
<Armbian-Discord> <I​gorPec> @cats beta is still working on old ?
<Armbian-Discord> <c​ats> yeah
<Armbian-Discord> <c​ats> I'll swap it soon
<Armbian-Discord> <I​gorPec> how soon?
<Armbian-Discord> <I​gorPec> just need to know if i have to enforce another workaround to rebuild caches
<Armbian-Discord> <c​ats> I can swap it in a few minutes? Just in a game atm :)
<Armbian-Discord> <I​gorPec> when you find time, no rush
<Armbian-Discord> <c​ats> seems like beta.armbian.com forces https?
<Armbian-Discord> <I​gorPec> yes
<Armbian-Discord> <I​gorPec> imola is not forcing https anymore
<Armbian-Discord> <c​ats> nice
califax has quit [Remote host closed the connection]
califax has joined #armbian
SallyAhaj has quit [Remote host closed the connection]
<Armbian-Discord> <c​ats> IgorPec swapped both archive and beta to the new redirector, seems to have fixed the http issue too
samythemany1 has joined #armbian
samythemany11 has joined #armbian
samythemany has quit [Ping timeout: 240 seconds]
samythemany1 has quit [Ping timeout: 240 seconds]
alekksander has joined #armbian
samythemany11 has quit [Remote host closed the connection]
alekksander has quit [Quit: Konversation terminated!]
<Armbian-Discord> <M​yron> An observation. On this here Discord under CHIP-SETS, should the A20 also be on the list?
xoan1 has joined #armbian
xoan has quit [Ping timeout: 246 seconds]
xoan1 is now known as xoan
archetech_n2 has quit [Quit: Konversation terminated!]
SallyAhaj has joined #armbian
SallyAhaj has quit [Quit: SallyAhaj]
SallyAhaj has joined #armbian
aasami_ has quit [Quit: good morning]