<div dir="ltr">That's "dropping off the wifi" not "dropping of".<div><br></div><div>Matt</div><div><br></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Dec 3, 2013 at 5:08 PM, Matthew Dalton <span dir="ltr"><<a href="mailto:msdalton@gmail.com" target="_blank">msdalton@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Hi Holger,<div><br></div><div>Thanks for the reply.</div><div><br></div><div>I'll try the git HEAD as a last resort, but for now I'm trying to figure out what's wrong while keeping the configuration constant.</div>
<div><br></div><div>Since my original post I have managed to get roaming working... but I did it by putting system("wpa_cli roam bssid") calls in my code. Clearly bgscan-controlled roaming could work (ie. there's nothing in my network setup to prevent it), but it doesn't seem to work for me.</div>
<div><br></div><div>Anyway, I'm having much bigger issues now, with clients randomly dropping of the wifi for no apparent reason. Putting "-ddddddt" in the command-line for wpa_supplicant doesn't produce anything useful. Mostly a spamming of STATUS and SCAN_RESULTS and the odd ROAM due to my forced roaming system calls. Nothing to indicate that the network was absent for a few minutes. I suspect the wifi access points are to blame for this, that their mesh/bridge implementation is flaky, but I'm still trying to find real proof that that's the case. If anyone knows where in the wpa_supplicant source code such errors may be detectable, please let me know, even if it's a reference to the current git HEAD of wpa_supplicant.</div>
<div><br></div><div>Cheers,</div><div>Matt</div><div><br></div></div><div class="HOEnZb"><div class="h5"><div class="gmail_extra"><br><br><div class="gmail_quote">On Mon, Dec 2, 2013 at 9:37 PM, Holger Schurig <span dir="ltr"><<a href="mailto:holgerschurig@gmail.com" target="_blank">holgerschurig@gmail.com</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">Hi Matt,<br>
<br>
most programmers here in this list use wpa_supplicant from the current<br>
git HEAD. So if you can compile that, replicate your problem and<br>
provide a log file, this would be helpful.<br>
<br>
You can git clone the hostapd project, then compile only<br>
wpa_supplicant. Then you can do "ifdown wlan0", and start the locally<br>
compiled wpa_supplicant: "./wpa_supplicant -s -i wlan0 -D nl80211 -C<br>
/etc/wpa_supplicant.conf -d". That way your Debian packaged<br>
wpa_supplicant won't be clobbered for the first test(s).<br>
wpa_supplicant automatically brings up the interface (equivalent of<br>
"ifconfig up", not setting IP adresses etc). But that is good enought<br>
for roaming tests.<br>
<br>
Also note that wpa_cli has a command "roam <bssid>" as a debugging<br>
aid. So you can try to roam manually to the AP in question, and then<br>
observe the debug output.<br>
<br>
Maybe the AP is simply sending beacons that he needs encryption, you<br>
wpa_supplicant.conf file doesn't contain any encryption, so APs with<br>
demand that are ignored for roaming decisions.<br>
</blockquote></div><br></div>
</div></div></blockquote></div><br></div>