<stintel>
Tusker: how did you request the sources from WG? they transfered ownership but now I can't open a case because no support contract
<Tusker>
stintel: 1) email legal@watchguard.com 2) create a "customer care" support ticket
<Tusker>
in both 1 and 2, mention the other activity...
<stintel>
k
<stintel>
something totally different: does anyone know how to register an object in ubus via shell script?
<stintel>
I would like to use ubus wait_for until something is done executing before continuing in another part
<stintel>
but yeah that seems impossible
<Tusker>
Slimey: is that the factory jffs2 partition? did you include the CONFIG_JFFS2_FS_WRITEBUFFER CONFIG_JFFS2_RTIME or CONFIG_JFFS2_ZLIB options ?
<Tusker>
if you boot into initramfs, can you mount that jffs2 manually ?
<Tusker>
otherwise, are you sure that the jffs2 partition is at that location ?
<Slimey>
its a clone of the PowerCloud Systems CAP324
<Slimey>
but adtrans firmware
<mangix>
a lot of flash and ram
<Slimey>
the ap's are fat and do all the hardwork but are managed via a "controller"
<mangix>
interesting. why is the ethernet interface hooked up to the switch?
swalker has quit [Remote host closed the connection]
<Slimey>
note sure
swalker has joined #openwrt-devel
<mangix>
weird
<aparcar[m]>
rsalvaterra: ping
<aparcar[m]>
do you use UML?
<digitalcircuit>
slh: Got my remote NBG6817 working again, it has the same MMC identifiers as my local NBG6817. Though I did not have the time/patience to test it as well (both routers are at families' homes), I suspect this means anyone with a 2018-ish or newer NBG6817 will experience a TFTP-recovery-required "soft" brick upon flashing any build with Linux kernel 5.10.
<digitalcircuit>
(I can see why it'd be useful to find an external hardware-based reset, e.g. 3 failed boots. At some point maybe I can disassemble the ZyXEL bootloader.)
<slh>
grr
<slh>
you have the GPL source?
<digitalcircuit>
slh: Nope. I didn't even think about checking ZyXEL for that; that makes a LOT more sense :)
<rmilecki>
jow: I don't want another release pain like with DSA
<rmilecki>
jow: so I really care about your opinion on whether firewall4 is ready enough
<rmilecki>
aparcar[m]: ^^
<aparcar[m]>
rmilecki: you don't think 6 month is enough?
<rmilecki>
aparcar[m]: i think jow may know
<rmilecki>
aparcar[m]: if noone works on it, a year won't be enough
<rmilecki>
aparcar[m]: when it came to LuCI and DSA so many people cared but very few actually worked on it
<aparcar[m]>
rmilecki: agree. Hopefully people start looking at it and find things that need work
<aparcar[m]>
*and work in things once they found things to work on
<mangix>
rmilecki: what was the pain in DSA?
<aparcar[m]>
mangix: massive delay of 21.x due to complex changes in luci and underlying config
<mangix>
well, do note there were devices using DSA even with 18.06
<aparcar[m]>
but compatible with luci?
<mangix>
I don't see why not.
<digitalcircuit>
aparcar[m]: Since you're listed as the contact email, would you be the right one to ping about working around the Let's Encrypt cert for https://sysupgrade.openwrt.org for the "auc" client?
<digitalcircuit>
On 21.02.0, it works with luci-app-attendedsysupgrade, but "auc" fails ("Connection error: Invalid SSL certificate"). Earlier, jow mentioned removing the expired backwards-compat cert chain for https://downloads.openwrt.org/ to fix similar issues (visible when comparing with openssl or https://chainchecker.certifytheweb.com/ ).
<stintel>
trying uxc but getting a bunch of: Wed Oct 6 05:06:19 2021 user.err : jail: can't open /sys/fs/cgroup/cgroup.subtree_control: No such file or directory
<rmilecki>
mangix: LuCI didn't have support for "device" UCI sections which are very helpful for DSA setup
<stintel>
looks like this is expecting cgroup2 but cgroupfs-mount is mounting v1?
<rmilecki>
also bridging support was /asking/ for some improvements
<rmilecki>
i've tried that and pushed some cool LuCI changes which broke a bunch of stuff
<rmilecki>
then jow had to help me sort all regressions out
<mangix>
rmilecki: got it
<stintel>
and why is uxc trying to mount extroot
<aparcar[m]>
stintel: ask dangowrt
<mangix>
json_add_int mtu "${mtu:-1280}"
<mangix>
wonder why so low
<aparcar[m]>
digitalcircuit: I’m using caddy to reduce my involvement in certificate managing but it still causes issues… I’ll look at it later. Please use LuCI in the meantime
<digitalcircuit>
aparcar[m]: Sure thing, no rush! I don't mind using LuCI, I just wasn't sure if this was a known issue or overlooked - you've clarified it's the former.
<aparcar[m]>
digitalcircuit: thanks for reporting
<mangix>
so 1280 is the minimum MTU. lovely.
clandmeter has joined #openwrt-devel
<digitalcircuit>
aparcar[m]: And thank you for your work on attended-sysupgrades - I haven't gotten to try it on a stable release yet, but it should simplify managing a few routers for family members.
ecloud has quit [Ping timeout: 480 seconds]
<rsalvaterra>
aparcar[m]: Sleepy pong. UML? The modelling language, or user-mode Linux?
ecloud has joined #openwrt-devel
rsalvaterra has quit [Quit: rsalvaterra]
decke has joined #openwrt-devel
rsalvaterra has joined #openwrt-devel
<stintel>
blocktrron: where's poemgr? commit message for usw flex says packages feed but don't see it ?
<stintel>
PaulFertser: thanks, had found it already, a bit sad that a commit in openwrt.git claims it's in the packages feed though
<stintel>
apparently today is new goodies arrival day =)
<stintel>
2x Unifi Switch Flex just arrived and 2x Edge-Core ECS4100-12PH loaded on the delivery vehicle
<blocktrron>
stintel: I still have to add the package to the repo
<blocktrron>
in the meantime, you can add the poemgr git as a feed
<stintel>
blocktrron: please do so :)
<stintel>
I've added it locally to my old lede-wip feed
<rsalvaterra>
stintel: Are you building your own internet? :)
<stintel>
interesting
<stintel>
make[3]: *** No rule to make target '/home/stijn/Development/OpenWrt/openwrt/build_dir/target-mipsel_24kc_musl/poemgr-2021-05-16/.prepared_ecb76ad520d2e8b19ddb21d62cf6a4ee_6664517399ebbbc92a37c5bb081b5c53', needed by '/home/stijn/Development/OpenWrt/openwrt/build_dir/target-mipsel_24kc_musl/poemgr-2021-05-16/.configured_68b329da9893e34099c7d8ad5cb9c940'. Stop.
<stintel>
rsalvaterra: nah, the flex are for me so I can ditch a PoE injector and add an AP upstairs, finally
<stintel>
rsalvaterra: the edgecore are from client, for testing OpenWiFi
<rsalvaterra>
Hm, makefile shenanigans?
<stintel>
maybe I didn't look hard enough and I cannot just drop the Makefile in a feed
pmelange has joined #openwrt-devel
<stintel>
blocktrron: if I install poemgr.git as a feed I still get the above error
robin_ has joined #openwrt-devel
danitool has joined #openwrt-devel
<stintel>
blocktrron: sorry, disregard
* stintel
runs to a shop to get coffee
<blocktrron>
:D
<stintel>
it was obviously still in my own feed so it picked it up from there again
<stintel>
pretty excited to start using this thing though
<stintel>
will require me to rethink my AP placement though
rmilecki has quit [Quit: Konversation terminated!]
rmilecki has joined #openwrt-devel
<stintel>
hehe good luck
dangole has joined #openwrt-devel
pmelange has quit [Quit: Leaving.]
rejoicetreat has joined #openwrt-devel
rejoicetreat has quit []
danitool has quit [Quit: Cubum autem in duos cubos, aut quadratoquadratum in duos quadratoquadratos]
rsalvaterra has quit [Quit: rsalvaterra]
rsalvaterra has joined #openwrt-devel
<stintel>
blocktrron: fyi, I had power-budget-exceeded when I enabled PoE on 4 ports but with only one connected
<stintel>
is that expected?
valku has quit [Quit: valku]
rua has joined #openwrt-devel
<stintel>
blocktrron: actually that happens even just when enabling
jlsalvador has quit [Read error: Connection reset by peer]
<stintel>
interesting, all 4 ports show enabled even though I've only enabled 1 in /etc/config/poemgr
<stintel>
and restarting the service causes the power-budget-exceeded error again
<stintel>
also, for the DSA people here, is there a way to apply a vlan config change for a single vlan, or a single port?
<stintel>
bringing down all traffic on a switch for something like that is really not acceptable :)
jlsalvador has joined #openwrt-devel
<stintel>
yeah that poemgr does seemingly random things
<stintel>
manually running poemgr apply seems to apply the uci config, but restarting the service messes things up
nitroshift has quit [Quit: Gone that way --->]
jlsalvador has quit [Read error: Connection reset by peer]
jbowen has quit [Ping timeout: 480 seconds]
jlsalvador has joined #openwrt-devel
jbowen has joined #openwrt-devel
jbowen has quit [Ping timeout: 480 seconds]
jbowen has joined #openwrt-devel
jbowen has quit [Ping timeout: 480 seconds]
jbowen has joined #openwrt-devel
jbowen has quit [Ping timeout: 480 seconds]
jlsalvador has quit [Quit: jlsalvador]
jbowen has joined #openwrt-devel
cp-- has joined #openwrt-devel
cp- has quit [Ping timeout: 480 seconds]
jbowen has quit [Ping timeout: 480 seconds]
danitool has joined #openwrt-devel
hanetzer has quit [Ping timeout: 480 seconds]
cp-- has quit [Remote host closed the connection]
<stintel>
and the PoS macbook requires another reboot because no more internet
<stintel>
horrendous junk
<SwedeMike>
stintel: sure you're holding it right?
<rsalvaterra>
stintel: The best Mac ever is the Power Mac G5 (the PCIe model). :)
<stintel>
macOS is junk. I've seen the same problem on different macbooks on different macOS versions with years apart, on different networks, in different companies
<rsalvaterra>
Mostly because everything is IBM inside. ;)
<stintel>
there, I've requested a corp Linux ThinkPad
<SwedeMike>
stintel: I run Linux VMs on my iMac. Best of both worlds.
hanetzer has joined #openwrt-devel
<stintel>
there's nothing good in the mac world for me
<stintel>
only frustration due to neverending breakage
<stintel>
so the alternative is a thinkpad with fedora
jbowen has joined #openwrt-devel
dangole_ has joined #openwrt-devel
dangole has quit [Remote host closed the connection]
cp- has joined #openwrt-devel
danitool has quit [Read error: Connection reset by peer]
danitool has joined #openwrt-devel
<blocktrron>
stintel: mind sharing your config?
<blocktrron>
the service is kind of messy, i know that
<svanheule>
stintel: why is the power budget so high for each PSE port?
<svanheule>
stintel: I know some Realtek switches can use either the actual power usage or the allocated budget to determine wether or not the device is overloaded
<svanheule>
if you're using "allocated power" instead of "real power", I could understand why it errors out the moment you enable a port
<svanheule>
but I also don't know the specific of this PSE chip, so I'm just guessing and blocktrron will have to give you the actual explanation :)
<mangix>
rsalvaterra: do your ath79 switch LEDs work with DSA?
<mangix>
stintel: too expensive
<mangix>
hmm. is this an ARM platform?
<rsalvaterra>
mangix: Haven't really paid attention to the LEDs. IIRC, they're working on the TL-WDR3600 @GbE. Don't ask me about the Archer C6. :P
<mangix>
that one was without LEDs, right?
<rsalvaterra>
mangix: Ah, right, the C6 LEDs are connected to GPIO, nothing to do with the switch.
<mangix>
right. so according to Ansuel, if no LED rules are specified, the default ones are used. Which begs the question on whether or not they should be specified.
<rsalvaterra>
My gut feeling would be just to leave them at their defaults.
<mangix>
mine is too.
<mangix>
OEMs make interesting decisions sometimes
<rsalvaterra>
You mean "interesting" as in Terry Pratchett's "Interesting Times", right?
<mangix>
questionable
<stintel>
svanheule: I'm not using anything, I just enabled the ports
<mangix>
there's a xiaomi device that has a hilariously bad switch config.
<stintel>
svanheule: everything else you see is poemger
<mangix>
says it
<svanheule>
stintel: sorry, I didn't scroll down far enough apparently :-/
<mangix>
's 8327 when it's actually 8337, references eth1 when there is no such thing. references it on port 5 when it should be port 6, set 3 LEDs per port when there are only two.
<mangix>
a complete joke basically.
<svanheule>
stintel: I don't know if the two behaviours I described are available on the PD69104, so maybe just ignore what I said
<stintel>
svanheule: I don't know anything about PoE on this level
<stintel>
I got 2 USW-Flex because they address a longtime need, and I just want to make things work :)
<stintel>
and I managed to crash my ECW5211
<stintel>
I blame qsdk kernel
ynezz has quit [Remote host closed the connection]
<rsalvaterra>
Seems sensible. When in doubt, blame Qualcomm. :)
<stintel>
for once it was me :P
<stintel>
and of course I'm testing on the one testing device without RJ45 serial console
minimal has quit []
KGB-0 has quit [Ping timeout: 480 seconds]
KGB-0 has joined #openwrt-devel
<aparcar[m]>
rsalvaterra: experience with fw4?
<aparcar[m]>
digitalcircuit: ping
<aparcar[m]>
digitalcircuit: I modified the caddyfile and created new certificates, it should work again with auc (at least in my tests)
<rsalvaterra>
aparcar[m]: Zero. About the same as the nftables it's based on.
<rsalvaterra>
:P
<aparcar[m]>
rsalvaterra: bueno
<rsalvaterra>
That will change, eventually, of course.
pmelange has joined #openwrt-devel
<digitalcircuit>
aparcar[m]: Thank you! This works great, auc can connect on both NBG6817s I have (both on 21.02.0). I didn't realize "auc" offered to flash if packages are updated, too; that's a nice touch (LuCI Attended sysupgrade showed me as being on the latest). No need to do that right now, but it'll be useful in the future.
<digitalcircuit>
I appreciate your time spent looking into this, too :)
<aparcar[m]>
digitalcircuit: I'm thinking of extending the LuCI view in the near future to show package updates
<aparcar[m]>
digitalcircuit: well, please tell your friends to use it and report bugs
<aparcar[m]>
rsalvaterra: you should also test it btw
<rsalvaterra>
aparcar[m]: LuCI? Or fw4?
<digitalcircuit>
aparcar[m]: Sure thing! I'll give the auc upgrade a try soon, just to see how it all works. OpenWRT's updating process had felt a bit lacking to me for less-technical friends/family, which this helps a lot, and if I find anyone asking around, I'll suggest it.
<aparcar[m]>
rsalvaterra: sysupgrade.openwrt.org
rmilecki has quit [Ping timeout: 480 seconds]
Tapper has quit [Quit: Tapper]
<stintel>
can that thing accept a /etc/build.config file and generate images with the exact same config?
<aparcar[m]>
stintel: no it's using the imagebuilder so individual kernels are currently not supported
<aparcar[m]>
it would take a bit to much CPU power, however I'm thinking of allowing different kernel styles if provided externally, e.g. SELinux
<aparcar[m]>
mangix: what's stopping this ccache update?
<stintel>
alright, can't really test it then
<aparcar[m]>
stintel: mind elaborating on your changes?
<rsalvaterra>
aparcar[m]: With auc, right?
<aparcar[m]>
rsalvaterra: I figured you'd prefer auc over LuCI
<rsalvaterra>
O:)
<stintel>
aparcar[m]: my images are so custom I don't know where to begin
<stintel>
and maybe 15% of my OpenWrt devices is even running targets that don't exist upstream or devices that aren't supported in an existing target
<aparcar[m]>
stintel: seems like you should collaborate more with upstream
<aparcar[m]>
hide
<stintel>
aparcar[m]: sure, if you pay me at my normal rate for the time spent on it, you can demand that
<aparcar[m]>
stintel: wait you don't want things to be upstream?
<stintel>
also, all of it is public, most of it either sent out as patch series once, or thread in the forum
<stintel>
I do, but I also have a life and a job
<aparcar[m]>
🙂
<stintel>
simply not a nice remark, I spend *a lot* of my time on this project
<aparcar[m]>
oh I hope is was clearly a joke since you're a project member
<aparcar[m]>
Allowing custom kernels may work using a precompiled toolchain. I'll see how fast this would work
<aparcar[m]>
Imagebuilders take on the current server like 30 seconds to build an image, with custom kernels I'm guessing it's more like 15 minutes
<stintel>
probably not worth investing too much time in that anyway, don't know if many people would use it
<aparcar[m]>
rsalvaterra: regarding yesterday, I've seen you touched UML, do you use and compile it?
<aparcar[m]>
stintel: I'm using auc for device testing of latest snapshots so I could see that it's useful for devs. If you have a fleet of custom devices with custom packages and kernel options you could update them all within minutes (i.e. 15)
<aparcar[m]>
Anyway long term thing, if you have a default kernels/targets running please give it a try and please don't be discouraged by my faulty comment
<stintel>
some of those need work, some of those dunno if still relevant, some of those not really ready for pushing upstream
<stintel>
that reminds me, I should really invest some time in vallumd to make a new release
jbowen has quit [Quit: leaving]
<aparcar[m]>
Yea external package feeds would be interesting to offer but likely also a bit of a security issue. I need containers if allowing that
<rsalvaterra>
aparcar[m]: I touched UML? Not intentionally, for sure. o_O
<rsalvaterra>
Commit?
<aparcar[m]>
rsalvaterra: nevermind, I thought 3326b5e75c277b4fac21bffd2085df4aa40d2775 but it's treewide
<slh>
and even with containers it would be a huge risk/ burden
<stintel>
ughh I really need to get more handson experience with kubernetes
<stintel>
everytime I'm using it again I don't remember anything
pmelange has quit [Quit: Leaving.]
<rsalvaterra>
aparcar[m]: Is there any documentation for attended sysupgrades, auc and the LuCI app?
<aparcar[m]>
rsalvaterra: GitHub
<rsalvaterra>
Thanks, will take a look.
Tusker has joined #openwrt-devel
<rsalvaterra>
aparcar[m]: Anything specific you need to test? Or just use the thing and see if it craps out? :)
goliath has quit [Quit: SIGSEGV]
<aparcar[m]>
rsalvaterra: if you have a bunch of devices and want them to use the latest snapshot daily, it can come in handy. That's at least how I use it
<rsalvaterra>
aparcar[m]: I'm discussing this with a friend of mine, who maintains several dozens of OpenWrt devices in production. This feature comes really handy to him.
<rsalvaterra>
Hopefully I can goad him into testing it… :P
<aparcar[m]>
yey
<stintel>
even my 2 newest devices with hastily created config I cannot use the auc for :(
<stintel>
because poemgr is not in core or feeds
<Tusker>
can't you setup your own auc build that includes your feeds ?
<stintel>
probably
<aparcar[m]>
Tusker: yes. You can also send me your binary feeds and I add them to my testing server
<aparcar[m]>
When is poemgr added?
<stintel>
well there was a thread about it on the ML a few weeks ago, I feel the responses might have been a bit discouraging
<Tusker>
currently it only supported the pd69104 ?
<stintel>
took me less than 12h after delivery to put 2 of those Unifi Switch Flex in production
<stintel>
Tusker: the only thing that matters to me is "currently allows me to control PoE on the Unifi Switch Flex"
<Tusker>
:)
Ansuel has joined #openwrt-devel
<rsalvaterra>
ldir: I'm looking at https://git.openwrt.org/?p=openwrt/openwrt.git;a=commitdiff;h=ce793584e52c9cf96d71344fd3515da9474c376f
<rsalvaterra>
Is there any reason for queued spinlocks to be disabled on PowerPC?
<stintel>
why you ask ldir? :)
<rsalvaterra>
stintel: Do you have queued spinlocks disabled on your M300?
<stintel>
let me check
<rsalvaterra>
stintel: Because he has a TL-WDR4900. :)
<stintel>
target/linux/qoriq/config-5.10:# CONFIG_PPC_QUEUED_SPINLOCKS is not set
<stintel>
probably default
<rsalvaterra>
4c/8t… you should probably enable them for better performance.
<stintel>
I can give it a try
<stintel>
but ... maybe next month :(
<stintel>
tomorrow I'm having dinner party, Friday bed early, Saturday morning to hometown for several weeks, so I won't be able to test as much :(
<stintel>
my testing M300 is unplugged in favor of the new ECS4100 switches
<rsalvaterra>
Oh, nice! Enjoy! :)
<stintel>
not looking forward to the 2100km drive
<stintel>
even if I'll pass autobahn
<rsalvaterra>
Are you driving alone?
<stintel>
yeah
<rsalvaterra>
From Bulgaria to the Netherlands…? o_O
<rsalvaterra>
Ouch… take care. Don't push it and always take a rest when you feel the need it
<stintel>
yeah, I've been doing this for 6 years now, I know the drill. but thanks for the concern ;)
<stintel>
Bulgaria to Belgium* but not that big a difference
<stintel>
my personal record is 17h :D
<rsalvaterra>
Always obeying the speed limits, of course… :P
<stintel>
more or less
<stintel>
usually +20 above it
<stintel>
but can't obey if there is none ;)
<rsalvaterra>
My biggest driving scare I ever had was exactly on a long trip, driving at late night. I opened my eyes… but I couldn't remember when I had *closed* them.
<stintel>
yeah, thin line to cross really
<rsalvaterra>
At that point I just stopped at the next petrol station and slept for a couple of hours.
<stintel>
yep, I'm pretty good at detecting the almost point, take the next parking lot or petrol station and sleep until wake up and go again
<stintel>
sometimes that's 15 minutes, sometimes that's 5h
<stintel>
it's probably related to how many nights I've been going at it until after 4 in the morning that week :P
<stintel>
so I'm actually considering to build a moveable homelab with those edgecore switches
<stintel>
problem is that I'll pass Serbia, non-EU, and they might give me a hard time about a lot of hardware in the car
<stintel>
so I'm going to limit it this time
<stintel>
but planning to spend winter in Crete or so and take a bunch of stuff there so I can comfortably work and hobby from there
<rsalvaterra>
Nice! Let me know if/when you come Portugal. ;)
<rsalvaterra>
Off to bed now. Over and out. :)
<stintel>
I'll definitely ping you when I'm coming that direction
<stintel>
nn
<stintel>
I should do the same, lights out in 1m :P