00:52 <philipp64> Seeing this failure about isc-dhcp-server-{ipv4,ipv6} both trying to install /etc/init.d/dhcpd ... I don't remember seeing that. Is that a new thing? Do I need to restructure the package as "variants"? https://github.com/openwrt/packages/actions/runs/4305395582/jobs/7507772879

00:57 <philipp64> Actually, everything that needs to be variant already is, so I'm not sure why the script checker is losing its mind...

03:36 <hgl> In a package's postinst, what would be a good way to add a cron job that will also work for the image builder?

03:44 <dwfreed> IPKG_INSTROOT is set in the image builder

03:44 <dwfreed> to where the target fs will be made from

05:22 <hgl> dwfreed: If I simply do "echo ... >$IPKG_INSTROOT/etc/crontabs/root" in postinst, will that work? Will the created image write to that file after the image is flashed and booted?

05:22 <dwfreed> yes, it's empty during normal usage

05:23 <hgl> from my research, the image builder seems to run postinst at build time instead of runtime?

05:23 <dwfreed> yes

05:25 <hgl> so that command in postinst causes the image builder to create the file in the image, which in turn gets flashed and the file created, is that correct?

05:26 <hgl> If I need to remove the cron job upon uninstalling the package, I guess there is no need to use IPKG_INSTROOT in postrm, right?

07:16 <damex> i have multiple ipq40xx + multiple mt7622 doing 802.11r roaming/fast transition. if management frame protection is enabled - no roaming happen between access points. i made experiment with two exact aps - one with management frame optional and one is required. one that had optional - worked fine. does not matter which client. multiple windows/linux laptops with intel, macs with broadcoms, recent iphones with whatever is in there, consoles and

07:16 <damex> whatever else

07:16 <damex> is it expected or i better dig deeper?

13:14 <djfe_> damex: I can't help, but CT or qcom wifi driver for the ipq?

16:40 <damex> djfe_: CT i suppose. 'kmod-ath10k-ct-smallbuffers' replacing default kmod-ath10k-ct

16:41 <stintel> damex: I've not noticed any roaming issues with MFP

16:41 <damex> i tried ipq40xx-only devices and mt7622-only devices and it has the same behaviour.

16:42 <damex> stintel: mfp?

16:42 <stintel> Management Frame Protection ?

16:42 <stintel> that's not entirely true, 11r has some issues, but the clients roam, that part is not the problem

16:42 <stintel> it's after roaming they sometimes get denied assoc

16:44 <damex> hm... maybe missconfiguration? hm... wpa3-sae + ft over air + same mobility domain and choose Generate PMK locally on all aps. pretty much all that i do to make it run from the scratch

16:46 <PaulFertser> SAE + FT is a different cipher, do you see it in "Supported Ciphers" in iw list on the clients?

16:49 <PaulFertser> 00-0f-ac:9 AKM is needed for SAE FT

16:49 <damex> akm?

16:50 <PaulFertser> But I thought the question was about RSN + MFP, not SAE.

16:50 <PaulFertser> Authentication and Key Management

16:50 <damex> group cipher/pariwise cipher - CCMP, Authentication suite - SAE FT/SAE

16:51 <damex> linux clients see that

16:51 <PaulFertser> damex: so do your clients have 00-0f-ac:9 in the "iw list" output?

16:53 <damex> PaulFertser: no, what is it supposed to be?

16:53 <PaulFertser> damex: it's supposed to be an entry in the "Supported Ciphers" list. If it's not there then it likely means the client can't use FT with SAE.

16:58 <damex> PaulFertser: thanks , i got the idea. gotta check the clients. there is no 'supported ciphers' on that particular client - i will check and compare others first. gotta see if i could find list of supported ciphers on others too

17:37 <karlp> dwfreed: hgl_ might it be time for openwrt to have cron/runparts standard? so apps can drop files into a cron.d dir instead of having them all try and manage grepping and sedding a shared crontab?

23:34 <Lechu> tmn505: I got myself a few RB911-5Hn a day before I noticed your forward-port to ath79 and decided to take a look at that pesky user LED glow

23:35 <Lechu> I checked pinctrl settings based on my experience with Ruckus stuff, and tried disabling JTAG (user LED is on GPIO3, so TMS), and disabling clock observation, but - nope. Do you have a pinctrl status dump from ar71xx by any chance?

23:37 <Lechu> It seems that routerboot disables JTAG on it anyway.