This has been reported. See OpenWrt 23.05.4 - Service Release - #211 by patrakov
Note that the 5 GHz slowdown and the 6/12 Mbit/s issue seems to be device-specific: Netgear WAX202 is affected (and has been downgraded), while Linksys E8450 isn't.
This has been reported. See OpenWrt 23.05.4 - Service Release - #211 by patrakov
Note that the 5 GHz slowdown and the 6/12 Mbit/s issue seems to be device-specific: Netgear WAX202 is affected (and has been downgraded), while Linksys E8450 isn't.
Power saving is more likely added by monthly client upgrade, otherwise totally matches nss1 fixed in snapshot.
In any case - how do I restore, in a 23.05.x snapshot, the throughput (according to speedtest.net) of the WDS client/repeater that was achieved in 23.05.3?
Works well with a bunch of decommissioned TERRA Black Dwarf ( Model 3v900 ) as of the early 2010s equipped with 1GB RAM will last at least another decade.
Hi, I'm sorry for the delay, just got back from vacation. Here is the tcpdump log with .3 firmware: https://pastebin.com/r1wsjFKL
Up & running on Linksys WRT1900ACS-V2: many thanks to all the dev guys.
Nothing really jumps out here either, except maybe this: The DHCP requests from the printer in the .3 case seem to be not from a "cold" restart of the printer, because it is trying to obtain/renew its existing/previous IP address:
Requested-IP (50), length 4: 192.168.1.172
That field isn't there in the .4 case. It's a long shot, but can you try cold-rebooting the printer while the router is still on .3 firmware and see if it a) still gets an address and b) that it does in fact not request any specific address - so same as the .4 capture?
In my experiment I started tcpdump session and then turned printer on. So I think that should be a "cold" start? Please let me know if I'm wrong. I powered on printer earlier in the day so router could have non expired dhcp lease for it.
The difference is in the request packet sent from the printer so if there's some state preserved, that's on the printer side.
BTW I think this subthread should continue either on a separate thread on this forum or perhaps on a github issue in the OpenWrt project.
WildCarrot, is there any sign of the throughput issue described by caywin above (and following replies) or is it OK?
Maybe too early, but I easily see more than 400 Mbit/s; anyway I did not take notice of performances with 23.05.3, so I can't say if .4 is better or not...
I loaded 23.05.4 on a new [TP-Link Archer AX23 AX1800 v1.0 and everything appeared to work fine. I am able to log into OpenWRT and update the config, but ran into a problem with the 5Ghz signal / channel. If I leave the channel set to "Auto" it sets up to channel 36 and I'm able to see it just fine with Wifi Analyzer on my phone. However, if I manually change it to any other port number, the 5Ghz signal disappears and does not reappear on the new channel, nor at all. Setting it back to "Auto" restores the signal. The 2.4Ghz signal works fine for changing the channel. Any thoughts or feedback would be appreciated.
What channel are you manually selecting? Maybe it's a DFS channel and that usually takes at least a minute to be available if there aren't any radars around you.
Cheers!
I tried lower range 56 / 60 and upper range 132, all with the same results. I believe I waited longer than a minute, but will try again and see.
Those are DFS channels in the US and many other countries. It can take several minutes (10+).
Try channels above 144 or below 52.
EDIT: I was going to suggest enabling background scanning for radar, but that isn't available yet on stable. You'd need to switch to snapshot, which isn't recommended for various reasons.
Background radar scanning, at least on my GL-MT6000, immediately makes the DFS channels available for connection. However, it doesn't work on boot, so don't use it.
Incase anybody missed it, Hauke has posted a thread asking for testers and to report any bugs on latest snapshot for eventual 23.05.5 release. Link here
Let's try and make 23.05.5 as bug free as possible.
I spoke too soon: wifi became hard to connect and slower than a snail; PC log plenty of deauthenticating from .... by local choice (Reason: 3=DEAUTH_LEAVING)
. After a while no way to connect at all... went back to 23.05.3
did the tailscale package just disappear i know i install it 2 day ago sniff sniff
23.05.5 has been released.