<ynezz>
bah, it didn't lasted that long `main_ipq40xx/generic` failed -> fatal: unable to access 'https://git.openwrt.org/project/luci.git/': The requested URL returned error: 504
detonate has quit [Ping timeout: 480 seconds]
hackpascal has joined #openwrt-devel
<ynezz>
I'll look into Fastly now
damo22 has left #openwrt-devel [#openwrt-devel]
detonate has joined #openwrt-devel
detonate has quit [Read error: Connection reset by peer]
<aparcar[m]>
ynezz: "we increase your limits as you grow"
<ynezz>
:D
<aparcar[m]>
they are really nice
<aparcar[m]>
I'm short on time but they offered me to meet an engineer to improve our cache system in general. Will do that soonish
<ynezz>
yeah, if we switch git to the CDN, we should modify the gitweb interface to add their logo/thank you to every page
<aparcar[m]>
yup I guess
<aparcar[m]>
Does gitweb support that?
<ynezz>
I hope so, when I've more time, I would like to look into some cgit based container and prepare git.staging.openwrt.org with cgit based interface
<aparcar[m]>
i still wonder if we shouldn't move to codeberg.de
<ynezz>
sure, but seems like CDN is probably fixing the issues we're currently seeing, so lets use that and move on, focus on more important topics?
<schmars[m]>
ynezz: thanks for the info :) don't know yet if it's git itself, but it seems unlikely to be a ddos sympton, because in my buildbot setup all four worker locations (completely different cities) started to have the issue at the same time, against g.o.o and github.com, which got my attention
<schmars[m]>
have just never seen this particular behaviour during clones
<ynezz>
but probably to only solution, change to git.cdn.o.o, do the feed update and then switch back into git.o.o
skynet2 has joined #openwrt-devel
skynet2_ has joined #openwrt-devel
skynet2 has quit [Ping timeout: 480 seconds]
skynet2_ has quit []
skynet2 has joined #openwrt-devel
Sawzall has quit [Read error: Connection reset by peer]
skynet2_ has joined #openwrt-devel
Sawzall has joined #openwrt-devel
skynet2 has quit [Ping timeout: 480 seconds]
rua has quit [Ping timeout: 480 seconds]
rua has joined #openwrt-devel
dermoth_ has joined #openwrt-devel
dermoth has quit [Remote host closed the connection]
dermoth_ is now known as dermoth
rua has quit [Ping timeout: 480 seconds]
hexa- has quit [Quit: WeeChat 4.4.3]
hexa- has joined #openwrt-devel
rua has joined #openwrt-devel
goliath has quit [Quit: SIGSEGV]
<Slimey>
openwrt.org web page is slow
<jakllsch>
old news
<robimarko>
If it suddenly gets fast, then we will get worried :)
<schmars[m]>
it makes me a little angry how much of our time our AI saviours are wasting
<aparcar[m]>
schmars: motivates me to make the wiki a static page eventually
<ynezz>
CDN it?
<ynezz>
its mostly static content already
<schmars[m]>
wiki static webpage would be lovely. shove it all into git
<schmars[m]>
would make the hw db much easier to work with too
<schmars[m]>
and it would make it easy to mirror it for offline access
<ynezz>
but you would cut out a lot of contributions
<ynezz>
(making a Git skills mandatory for contributions)
<Habbie>
you can still have some webfrontend for it
<schmars[m]>
yeah that worries me too, otoh it seems like motivation to maintain the wiki installation seems low too :) and the quality of many articles is really low, sometimes no info is better than wrong info
<schmars[m]>
i care most about the hw db since i use it programmatically in our build scripts
<efahl>
I think a big part of the issue with wiki maintenance is that since the departure of tmomas, the barrier to entry is huge
<efahl>
to get a wiki account, you have to come over here to irc and request one via pm, which is daunting for a lot of forum-only users
<ynezz>
how was tmomas handling it?
<ynezz>
AFAIK bobafetthotmail was as well doing a great work with wiki maintenance
<efahl>
he had a pinned forum post where you could just request it directly, no irc involved for the requestor
<ynezz>
so lets re-create it?
<efahl>
The top level thread is still there, but no one on the forums has admin on the wiki to be able to add users
<efahl>
I've long thought we should just auto-create a wiki account for forum members above a certain trust level, but that would probably require someone who knows both forum and dokuwiki security/db stuff (not me :) )
<ynezz>
yeah, automation avoiding spammers would be the best
<ynezz>
anyway, that @wiki-account approach LGTM as well, we should resurrect it
<Habbie>
doesn't github login work today? somebody told me it did
<efahl>
I heard that, too, but didn't check it out
<efahl>
I don't think it does, I think dokuwiki has its own auth plugin that's sort of limited
<ynezz>
I'm using GH for login
<ynezz>
you probably need the user created first, by the admin, no?
<ynezz>
and spammers can buy plenty of GH logins I assume
<ynezz>
so it wouldn't solve the spam issue, probably the main reason for self-creation
<efahl>
yeah, I justed tried gh login and it's circular, you authorize with gh, but don't have a user account, so you login with gh, ...
<Habbie>
heh
minimal has joined #openwrt-devel
totkeks has joined #openwrt-devel
<enyc>
Hrrm; I'm not into code/git side of things at the moment; but I'm curious (a) if 24.10- has any abstractions to simplify configuring same ssid/security across different radios where you want that ... and (b) if TKIP encryption on WPA2-only has been deprecated as a default, etc. ?
<schmars[m]>
ynezz: about worker hostname resolution, git has config options to transparently rewrite the origin url
<Habbie>
the superuser.com link i posted has an example of that
ScrewDriver1337 has joined #openwrt-devel
<schmars[m]>
ah nice, thanks
<schmars[m]>
going down the git clone failure rabbit hole now, if you never hear of me again, my gravestone shall read "no price is too high for a green build"
<Habbie>
lol
<schmars[m]>
here's the reproducer: podman run -it --rm docker.io/library/alpine:3.18 sh -c 'apk add git ; while true ; do git clone https://github.com/openwrt/packages.git /root/packages ; rm -rf /root/packages ; done'
<ynezz>
schmars[m]: how many iterations does it need? :P
<ynezz>
-EUNABLETOREPRODUCE
<schmars[m]>
first failure on 10th iteration, now i'm waiting for the next, i'm maybe at iteration 30 or so now
<schmars[m]>
you are in .cz right? so you're also ending up at frankfurt github?
detonate has joined #openwrt-devel
<ynezz>
de-cix.fra.github.com
<schmars[m]>
another - the clone started very slow, like 20 KB/s, then picked up speed and eventually aborted
<ynezz>
you should probably add 'GIT_TRACE_PACKET=1 GIT_CURL_VERBOSE=1 GIT_TRACE=1 git clone ...'
<ynezz>
error: RPC failed; curl 92 HTTP/2 stream 5 was not closed cleanly: CANCEL (err 8)
<schmars[m]>
yep - similar error when i tried http.version HTTP/1.1 btw. trying alpine:edge now for latest git, i only tried alpine:3.18 before because the sdk gave me a segfault on edge
<ynezz>
so perhaps really some network issues and maybe tcpdump would be able to see them?
<ynezz>
here it basically clones forever at full line rate speed
<schmars[m]>
yeah... just curious that it started at the same time at 4 completely different worker locations, against 2 completely different git services
<schmars[m]>
more worker locations actually - add the openwrt buildbot workers to that :)
totkeks has quit [Quit: Leaving]
owrt-images-builds has quit [Quit: buildmaster reconfigured: bot disconnecting]
<efahl>
the ubus call system board call shows 'sipeed,lichee-rv-dock', but in profiles it's 'lichee,rv,dock'
<efahl>
in the patched dts, we see: compatible = "sipeed,lichee-rv-dock", "sipeed,lichee-rv",
<efahl>
or do I need to manually add to (for this one case) target/linux/d1/image/Makefile
<efahl>
a like like SUPPORTED_DEVICES := lichee-rv-dock sipeed,lichee-rv-dock
<efahl>
I think all the targets in that one specific Makefile show this same mismatch between on-device system board output and download-site profile name
sorinello has quit [Ping timeout: 480 seconds]
<schmars[m]>
it' coming only from SUPPORTED_DEVICES in the target makefiles i think
<efahl>
yeah, that's sort of what I was thinking, thanks for confirming
<schmars[m]>
on the running device, it's platform-specific where /tmp/sysinfo/board_name is coming from
<schmars[m]>
dunno i'm just grepping, started in json_add_image_info.py
<schmars[m]>
which what's incrementally writing profiles.json