Web lists-archives.com

Re: slow connections to non-bridged hostapd/dnsmasq wireless access point (was pointers to material...)

(Now that I'm awake...)

On Wed, Jun 14, 2017 at 2:18 AM, Joel Rees <joel.rees@xxxxxxxxx> wrote:
> On Tue, Jun 13, 2017 at 9:26 PM, Dan Ritter <dsr@xxxxxxxxxxxxxxxx> wrote:
>> On Tue, Jun 13, 2017 at 11:55:11AM +0900, Joel Rees wrote:
>>> On Tue, Jun 13, 2017 at 9:30 AM, Dan Ritter <dsr@xxxxxxxxxxxxxxxx> wrote:
>>> > On Mon, Jun 12, 2017 at 11:50:00AM +0900, Joel Rees wrote:
>>> >> (famous last words)
>>> >>
>>> >> On Sat, Jun 10, 2017 at 3:53 PM, Joel Rees <joel.rees@xxxxxxxxx> wrote:
>>> >> > I now have connection for both the wireless and the netbook that is acting
>>> >> > as the AP. I took out the bridge entirely, quit trying to play with
>>> >> > port forwarding,
>>> >> > just used dead simple setup. dnsmasq was the only missing piece, if I had
>>> >> > not been focusing on bridging.  Bridging is probably for the other direction.
>>> >> >
>>> >> > But the wireless is pretty slow, so I'm not sure I'm finished.
>>> >> >
>>> >> > I have to go take care of some family business, when I'm done I'll
>>> >> > post the details.
>>> >> >
>>> >> > But it's really pretty simply. I was just working too hard.
>>> >>
>>> >> But it's too slow to maintain a connection.
>>> >>
>>> >> After mucking around a bit, I haven't really come up with anything. So
>>> >> I'll post my
>>> >> configurations (names changed as usual):
>>> >
>>> > You have an eth0 network, a non-overlapping wlan0 network -- can
>>> > you characterize "too slow"?
>>> Actually, now that I dig in, it only seems to be routing dns traffic.
>>> That is, I tried to ping the wireless interface from my daughter's computer
>>> and got a network not reachable or down result of some sort. (I didn't
>>> write the exact message down, and my daughter's at school so I can't
>>> use her computer.)
>>> But my logs on the netbook that is running hostapd and dnsmasq show
>>> quite a bit of dns traffic, lots of queries returning successful results.
>>> > Things to check:
>>> >
>>> > - ping from your netbook to the outside world
>>> No problem there.
>>> > - ping from a wifi client through the netbook to the outside
>>> >   world
>>> That also got network not reachable or down or something.
>>> > if those work,
>>> >
>>> > - wget -O /dev/null http://speed.hetzner.de/100MB.bin
>>> > - and on a wifi client
>>> I tried that in bridge configuration just now and I'm only getting one
>>> percent every twenty to thirty seconds. The non-bridged configuration
>>> doesn't even read the mail to look at the url.
>> Right, that's bad.

Actually, it might be normal for this ancient Android tablet.

>> Try this:
>> allow-hotplug eth0
>> iface eth0 inet static
>> allow-hotplug wlan0
>> iface wlan0 inet static
>> auto br0
>> iface br0 inet static
>>     address
>>     netmask
>>     broadcast
>>     gateway
>>     dns-nameservers
>>     bridge_ports eth0 wlan0
>>     bridge_maxwait 1
>> Note that now both the eth0 and wlan0 interfaces will have the same
>> address
>> But for right now, ignore that: just test connectivity and speed on this
>> netbook while the bridge is in effect.
> Interesting. Without the alias on eth0, the netbook is now able to access
> the internet through the bridged eth0 port. This is beginning to make
> more sense to me.
> Speed on your 100Mb blob is reasonable, ETA of about two minutes
> and 40 seconds, which is close to the limit on my connection.
> Wireless devices are not able to log in. I'm not sure I got the changes
> to the dnsmasq and ntp configurations right for doing a subrange of
> what I have the wall router providing me.

I forgot the most important setting.

Setting hostapd to bridge mode solved the connection problem.

It's still running about 20 seconds per megabyte, which may be all I
want to expect from this ancient single-processor 32-bit ARM
Android 2.4 tablet with only 1G RAM and an 80%-90% full 8G internal

I'll try a connection on one of my kids' computers later.

>> If you can't get a good approximation of your internet
>> connection speed from this, there's more to track down.
>> -dsr-
> This is something I can play with after I wake up tomorrow. (It's about
> two in the morning here and I need to get some sleep.)

I have a job interview I need to prepare for, so I'll have to save the
playing with the parameters for later.

Again, thanks. I'll post some of my results so far, maybe today, in case
someone else wants to do this.

Joel Rees

One of these days I'll get someone to pay me
to design a language that combines the best of Forth and C.
Then I'll be able to leap wide instruction sets with a single #ifdef,
run faster than a speeding infinite loop with a #define,
and stop all integer size bugs with my bare cast.

More of my delusions: