I am indeed willing to do that, but i am unsure if i'm capable or knowledgeable enough to do so. Do you (or anyone else) have an insight of how such reporting works? What is needed (log files (which?), dmesg?, device information?)
I do have a more experience now with running ath10k. It seems 2.4ghz radio is rock solid. But i do , mostly once in an entire evening, see 5ghz wireless dropping (client is an win10 laptop with intel 7260hmw wireless nic). A quick disconnect and reconnect solves the issue , but i didnt have this issue with the ath10k-ct driver.
So now i can choose between the ath10k driver which handles 2,4ghz fine , or the ath10k-ct driver that is more stable with 5ghz.
For now, a single 5ghz disconnect per evening is something for me that is "doable", the frequent 2.4ghz disconnects with the ath10k-ct driver i cant live with (need my wireless radio (which uses 2.4ghz) for sanity reasons )
Could it help if i report these issues to the ath10k developers (as suggested) and/or contact candelatech/ath10k-ct developers with these issues? Maybe we can make this a more collective effort, with collecting al relevant data from ath10k and ath10k-ct users and hand these over to the developers?
i think a reasonable place to start is to share some of the same information from this thread (perhaps incl. a link to this thread) on the mailing list for the driver. they will ask follow up questions if they have any.
To be more specific, which driver do you mean, the ath10k or the ath10-ct driver? Both have their strong points and their weaknesses. Would be nice if someone could merge both drivers (and corresponding firmwares) taking the strong and stable 2.4ghz performance of ath10k and the fast and reliable 5ghz-ac from the ath10k-ct driver and candelatech firmware. I am aware that this is no walk in the park, and maybe even impossible.
Knowing that candelatech operates under an qualcomm nda makes things more difficult. I guess qualcomm wont release their firmware under gpl, what would make a huge difference.
Does anyone know if the ath10k and ath10k-ct driver developers are in contact with each other, or do they work completely seperate from each other?
-off topic: in the mean time, i'm looking at what ath11k could possibly give us, and if devices supported under ath11k would be a "better" choice instead of hoping the c2600 one day will run stable on both radio bands.
-off topic 2: after hearing felix fietkau talk very positively about mediatek , i'm trying to find out if there are mediatek based devices that are somehwat comparable in specifications to the c2600 and are available in western europe.
Today i've upgraded to 21.02.2. Seems there has been a lot of development going on. With 30 changes in wireless / common alone, 21.02.2 seems to be a big deal. And even better: after upgrading i've encountered no drops on the 2.4ghz radio. As it was pointed out to me, after upgrading i should be on ath10k-ct again. I'll keep using ath10k-ct if it keeps being this stable. Fingers crossed.
Well, it seems that 21.02.2 is a real gamechanger. When 2.4ghz broke between 19.07.8 and 21.02, things didn't look good for ath10k-ct. Now, after running 21.02.2 for a couple of weeks, i'm pretty optimistic again. I've had no problems on both 2.4ghz and 5ghz. I have no idea what changed / broke in 21.02 , but it seems that it is now fixed again and ath10k-ct runs pretty solid.
@anon89577378 As i'm running my c2600 as an AP, i have 80% of total memory available. 370mb of 466mb total. Would i gain anything from switching to ct-smallbuffers?
Well, this was fun as long as it lasted.... I'm running 21.02.3 for about a day now. My 2.4ghz internet radio worked flawlessly on 21.02.2, now with the .3 release it play's for about half an hour to an hour, and then completely loses wireless connection. I have to psysically reboot the device, after that it plays fine again for a certain amount of time, before it again loses the wireless connectivity. Going back to 21.02.2 for the time being.
Does anybody know if things have been altered from 21.02.2 to 21.02.3 that affect ipq8064 / ath10k-ct? 2.4ghz isnt as bad as it was on 21.02.1, but it isnt where it should be. Feels like we took a step back in reliability for the c2600 on 21.02.3
I am also having issues with it recently (since 21.02 actually). Sometimes it just drops internet for some devices (can be wired or wireless) and actually turning on a device with a wired connection recovers the internet connection.
I'm currently running 22.03.0 on my c2600 devices. The previous issues seem to be gone and/or have been resolved. The issues i had with my wireless radio (logitech ue smart radio) where caused by a combination of its crappy wireless nic and recently introduced wifi6 equipment at my neighbours. They dont play nice together. Issue was solved with a wireless ethernet bridge.
So please keep this in mind if you suspect the c2600 failing, it can be that there are other things going on that disrupt your devices connection.
In the last couple of releases there seem to have been multiple updates issued to the -ct driver. This has resulted (for me) in a relatively stable working c2600. For me , the c2600 makes a great access point, and seems to be pretty future proof (if the software keeps being at this level). I see no reason to upgrade to wifi6.
Experienced a lot of problems lately as well. If i look back for several months. Excellent wifi speed which all of the sudden collapsed. Tried everything, all drivers. What did the trick for me was to adjust fast roaming. Changed it to FT over DS and disable generate PMK locally. After that a stable connection. Not sure if it does the trick for anyone else but definitely worth a try.
I never used "fast roaming", i need to read into that. I'm currently running 21.03.1 (just over a day) and havent been aware of any issues, other then the ones that already existed and probably cant be resolved. I'm talking about mesh networking on ipq8064 and ath10k-ct with candela tech's own firmware. This is why @tmomas i was hoping to keep this topic alive, since it would be great ( but my hopes are deminished) that all features on the c2600 would be utilized. That includes mesh networking and nss cores activated by default.
I was searching and reading a little into this, and found a comment from a candelatech developer on rawmode for the ath10k-ct driver, wich supposedly is needed for mesh networking to work:
I'm unsure if at some point in time someone has developed ath10k-ct rawmode, but according to github, the issue is still open.