gargoyle-ispy 2015-May-20 05:57.torrent

[ektus]

I could replace the WR702N with some other access point to see if that one is the problem, but why should it be if simply re-starting the wifi reliably revives the connection?

First shot: Replace the WR702N with another one I've got lying around. 4.18.107 Build 121123 Rel.31456n with different MAC and SSID, same channel. The problem remains, the wifi link goes dead after a short time.


Regards
Ektus.

[ektus]

First shot: Replace the WR702N with another one I've got lying around. 4.18.107 Build 121123 Rel.31456n with different MAC and SSID, same channel. The problem remains, the wifi link goes dead after a short time.

Next shot: As OpenWRT et al are too big for the tiny WR702N, I tried at least updating the stock firmware from 121123 to 4.18.102 Build 141203 Rel.61146n but to no avail. The connection still gets unresponsive


Regards
Ektus.

[ektus]

As none of my tries have been successful so far and 1.6.2 didn't bring any improvement either, I've gone back up to 1.7.x from may 20th, 2015. So from now on, we're once again on topic here

Currently, the router is at a location where it can barely (well, approx 35% quality) see my test WR-702. It fails to establish a connection for longer than 2 seconds:

Code: Select all

Sat May 30 20:31:07 2015 kern.info kernel: [  145.030000] wlan0: authenticate with 64:70:02:6f:04:13
Sat May 30 20:31:07 2015 kern.info kernel: [  145.040000] wlan0: send auth to 64:70:02:6f:04:13 (try 1/3)
Sat May 30 20:31:07 2015 kern.info kernel: [  145.110000] wlan0: send auth to 64:70:02:6f:04:13 (try 2/3)
Sat May 30 20:31:07 2015 kern.info kernel: [  145.180000] wlan0: send auth to 64:70:02:6f:04:13 (try 3/3)
Sat May 30 20:31:07 2015 kern.info kernel: [  145.260000] wlan0: authentication with 64:70:02:6f:04:13 timed out
Sat May 30 20:31:12 2015 kern.info kernel: [  149.820000] wlan0: authenticate with 64:70:02:6f:04:13
Sat May 30 20:31:12 2015 kern.info kernel: [  149.830000] wlan0: direct probe to 64:70:02:6f:04:13 (try 1/3)
Sat May 30 20:31:12 2015 kern.info kernel: [  150.050000] wlan0: send auth to 64:70:02:6f:04:13 (try 2/3)
Sat May 30 20:31:12 2015 kern.info kernel: [  150.050000] wlan0: authenticated
Sat May 30 20:31:12 2015 kern.info kernel: [  150.070000] wlan0: associate with 64:70:02:6f:04:13 (try 1/3)
Sat May 30 20:31:12 2015 kern.info kernel: [  150.080000] wlan0: RX AssocResp from 64:70:02:6f:04:13 (capab=0x431 status=0 aid=1)
Sat May 30 20:31:12 2015 kern.info kernel: [  150.090000] wlan0: associated
Sat May 30 20:31:12 2015 daemon.notice netifd: Network device 'wlan0' link is up
Sat May 30 20:31:22 2015 kern.info kernel: [  160.200000] wlan0: deauthenticating from 64:70:02:6f:04:13 by local choice (Reason: 3=DEAUTH_LEAVING)
Sat May 30 20:31:22 2015 daemon.notice netifd: Network device 'wlan0' link is down

It may be related to the fact that channel 3 is quite busy there, but nevertheles, some 30% signal quality should be plenty. 1.6.2 was behaving the same in this location, so no reason to stick with it either.

I'm going to re-install the router at it's normal location and try the ping watchdog next.


Regards
Ektus.

[ektus]

I'm going to re-install the router at it's normal location and try the ping watchdog next.

Did that, and simply entered the command "wifi" in the "Script:" field. No need to program anything

It seems to work. I've set the watchdog to check the connection once a minute, and have seen it come back several times already. Will check logread in a couple hours, but I'd say problem circumvented


Regards
Ektus.

[Lantis]

Not particularly relevant to your problem but, why channel 3?
1, 6 or 11 are the non interference channels.

[ektus]

Not particularly relevant to your problem but, why channel 3?
1, 6 or 11 are the non interference channels.

Are they? I'm in Germany, where channels 1 through 13 are available. Every WLAN survey tool I've seen shows a picture like below, where the networks are shown symmetric to their base channel, and don't extend outside the channel range. This is with 20MHz bandwith, don't have experience with 40MHz.

Regards
Ektus.

InSSIder scan of current channel usage at my desk. Networks with same channel act as repeater, so must be on the same frequency. The entire channel range is shown as in use, and no overlaps.

wlan_channels.png (33.87 KiB) Viewed 7831 times

[Lantis]

yes absolutely they are.
http://upload.wikimedia.org/wikipedia/c ... 29.svg.png

When you select channel 1 it extends "below" the graph you have posted. but that is well within the standard. it is designed that way.

As you can see from this picture channels 1, 6 and 11 are designed to be non-interference due to the lack of overlap.
on channel 3 you are overlapping the space of both 1 and 6.
3 and 7 overlap a minimal amount but it does still cause interference.
Being that you are in a place that allows channel 13, you could use 3, 8 and 13 to the same effect

just something to keep in mind. do what you want.

[ektus]

To follow up on that one:

Will check logread in a couple hours, but I'd say problem circumvented

It has been running overnight, and the watchdog has been active several times. It looks like the connection is lost more often at times when I'm browsing the log files in the SSH session, as during night time there was only one event per hour or so:

Code: Select all

Sat May 30 22:08:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 22:16:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 22:29:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 22:38:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 22:48:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:00:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:08:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:10:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:20:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:37:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sat May 30 23:52:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 00:08:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 01:08:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 01:58:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 02:08:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 03:09:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 04:09:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 05:09:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 06:09:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 06:20:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 06:22:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 06:45:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script
Sun May 31 06:53:13 2015 user.notice /usr/lib/gargoyle/ping_watchdog.sh: Run custom script

Regards
Ektus.

[ektus]

yes absolutely they are.
http://upload.wikimedia.org/wikipedia/c ... 29.svg.png

When you select channel 1 it extends "below" the graph you have posted. but that is well within the standard. it is designed that way.

As you can see from this picture channels 1, 6 and 11 are designed to be non-interference due to the lack of overlap.
on channel 3 you are overlapping the space of both 1 and 6.
3 and 7 overlap a minimal amount but it does still cause interference.
Being that you are in a place that allows channel 13, you could use 3, 8 and 13 to the same effect

More like 1, 7 and 13 to achieve much wider gaps. But what for? I'm installing more ethernet cables to reduce the WLAN usage anyway
As soon as 40MHz channels come to play, it becomes impossible to have three non-overlapping channels, or am I wrong here? But we're way off topic now

Regards
Ektus.

[ektus]

It has been running overnight, and the watchdog has been active several times. It looks like the connection is lost more often at times when I'm browsing the log files in the SSH session, as during night time there was only one event per hour or so:

Looks like a pattern there. It's another couple hours later, and the watchdog has been restarting the wifi once an hour, every hour, always at the same time approx. 9 minutes past the full hour. It has issued additional wifi restarts whenever I was doing something on the SSH console (which has been open all the time, but idle). The link is idle, too, statistics on the WR702 shows only a couple hundred packages in total for the past two days.


Regards
Ektus.