TREX: wireless-network Messages: 25 Entries..

The DLR lidar tower and etherant (eantE or 14.37) has been returned and is placed under the Base trailer. The antenna will be detached and placed in the back of the base.

Poulos and Agee executed tests where the Central and Hot-film etherants were switched between HotWire and NIDS SSIDs. We also manually altered the elevation and azimuth of the high gain antenna. 

In all cases while occasional improvements to dB were found (as high as 18dB), the system dB always varied significantly and to low levels (down to 6-8dB). No preferential orientation could be systematically established.

It was noted that REAL and West continued to maintain 20dB+ levels.

The DLR antenna has been returned and we could try it on Central since it consistently returned 20dB+ as well, but given the late state of the experiment we will likely be satisfied with reboots of the Central etherant 12.40 until the end.

Overnight the performance of the recently implemented eantJ 12.42 at the Central tower was flaky. Although initially a steady performing antenna with between 14-16dB, this system gradually reduced in effectiveness and was found at ~10dB this morning.

The plan then was to move this antenna to the high-gain antenna, which, with little hot-film traffic would be relatively free for Central. In addition, the high-gain antenna could increase the SNR of the new Central antenna by as much as 6dB ideally. With John's help Greg implemented new SSID that was incorrect and shutdown the wireless link to Central. We used 'hotfilm' instead of the actual name 'HotWire' and rebooted unsuccessfully.

As a result the central tower was visited and the ToughBook brought up into the ethernet board, which showed that 12.42 could not be pinged. An attempt was made to try a ping from the Central NIDS to no avail. The system was hardbooted in an attempt to recover functionality, unsuccessfully. As a result some data was entirely lost (roughly 45 minutes). 

Data began coming in again after switching the cable to the original antenna, 12.40, eantH, which had been left up on the tower disconnected when eantJ (12.42) was put up, just in case.

We found that 12.40 came right up with activity in all three lights, whereas reconnecting 12.42 there was 2 lights flickering and the wireless activity light not on. This indicates a wireless configuration problem, I suspect.

12.42 was detached and brought back to base. We do not plan to work on reconfiguring it here now since 12.40 is operating with 12-14dB currently.

We have not attempted to use HotWire as SSID with 12.40 yet.

23Apr06,
~17:45 Rebooted TPOP.  Reestablished link with central-eantJ, and sodar-eantg.

~18:30 Realigned very slightly eantJ at central.  Steven Mobbs, watched the
signal strength for us at the time and noted we got roughly 3db improvement.
Also slightly improved the tape job on the connector.  I didn't/couldn't
really pull apart the connector to see if water had gotten inside but my
impression is that it probably did not.

Written from Boulder...

central and west systems have not been reporting on the network for
the last few hours, but south is fine.  Looks like TPOP, which serves
192.168.12.0 is sick again.

west (192.168.13.71) is on the AP24 ethernet interface. The base has an 
etherant on both the .12 and .14 networks, so there are two ways to
route traffic to west, either through 192.168.12.41 (AP24 internal
antenna), or 192.168.14.1 (AP24 external antenna).

Up to this point we were routing though 192.168.12.41, which requires
the TPOP:

$ route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.14.0    *               255.255.255.0   U     0      0        0 eth1
192.168.13.0    ap24int         255.255.255.0   UG    0      0        0 eth0
192.168.12.0    *               255.255.255.0   U     0      0        0 eth0

I think it would improve things to route through the .14 network and
avoid the TPOP.

Changed the route using these commands:

route delete -net 192.168.13.0/24 gw 192.168.12.41
route add -net 192.168.13.0/24 gw 192.168.14.1

Also changed /etc/sysconfig/static-routes so that it takes effect on
the next reboot:

any net 192.168.13.0/24 gw 192.168.14.1


route command now shows:

$ route
Kernel IP routing table
Destination     Gateway         Genmask         Flags Metric Ref    Use Iface
192.168.14.0    *               255.255.255.0   U     0      0        0 eth1
192.168.13.0    ap24ext         255.255.255.0   UG    0      0        0 eth1
192.168.12.0    *               255.255.255.0   U     0      0        0 eth0

west is now responding.

On west, also changed the address for server in /etc/hosts
from 192.168.12.1 to 192.168.14.21.

In the /usr/local/isff/projects/TREX/ISFF/ops/ops*/ads3.xml, 
changed the destination address to 14.21:


     


On west, did /etc/init.d/nids restart and it is now sending data.

22Apr06, 8:30-9:30 PDT

The central etherant was down a couple times yesterday and was restarted twice
as late as midnight.  Then it was down again this morning at 5am.
Yesterday I pulled the eantj-sodar off the sodar to swap in at central.


When I climbed to central EantH, here are a couple observations while it
was still 'down.'
- power and wireless activity lights were present, looked good.
- the ethernet connection light was out.
	Note, Greg had wanted to go through swapping the pocketec proceduer
	before doing the antenna, and we had some troubles again
	with getting the swap-in to go.  However the ethernet should have
	been up, esp after a hard reboot of the viper.
	This could indicate a problem perhaps between antenna and
	viper instead of to tpop....maybe...
- the cable end going into the Eant, is improper because there is a gap
	between where the insulation was stripped back and the backshell
	of the connector.
	I taped this connector up speculating that it may be an outside
	possibility that our troubles began after the rain in late Mar,
	with perhaps some moisture getting in?....unlikely.
- 15vdc was OK.
Swapped in EantJ-sodar, installed just below the 'H'.   I pointed it slightly
differently, aiming a bit more down and a bit more to the north.....uncertain
whether this 'adjustment' might help or not.   It may move the pointing more to
the side lobe and perhaps away from interfering sources in town??
Left EantH installed in case it proves to be better than the new one.

Re: SNR seen from the Eants
From briefly monitoring the webpage for EantH yesterday we were saw ~13-15dbSNR,
which is marginal (ie 'yellow').  This afternoon while watching the EantJ,
we've been noticing anywhere from 9-17db reported from the antenna with
nominal around 12.

Re: Using EantJ-sodar instead of EantG-school at central.
I put EantG-school on the sodar yesterday afternoon instead of using it for
central.  The reason is because last summer while doing our range tests, 'G'
had been flakey and had an 'X' penciled on it.  During preparations for TREX
it did work OK in the staging area so decided to ship it hoping it would be
used for 'the other guys' or as a last resort.  Instead it found its way onto
the school, a critical link.   It's not out of the question that some of our
problems with 'School-network' were due to 'G' being flakey??
Meanwhile since yesterday afternoon it seems to be happy at the sodar site
200yds from the base.

John and I removed all the network equipment from the Independence
school yesterday.  That link was not being used and it gives us
a spare etherant WIFI antenna. 

The wireless network from central was down from about 9:25 - 10:35 PDT.
Steve and John restarted by cycling power on the etherant at central.

Wireless communication to central lost around 9 am PDT, restored around
1800 PDT.

Lost wireless connection to central around 0600 PST this morning.  Briefly lost
connection to west while Steve trying to fix this problem and dsm server and 
netcdf process stopped.  Fixed connection to west and restarted dsm server and
netcdf process.  Steve recycled power on the etherant at central to restart 
around 10 am PDT.

The link to central has gone down. This occurred at about 6:00 am this morning.
The link to the hotfilm is still good so I do not think it is the TPOP;
however, I will recycle power on TPOP to see if that fixes it.
Power recycle did not help. Time for visit to central.

REplaced 316 MHz MaxStream radio in logger with 2.4 GHz Zigbee radio.

Note that the intermittent logger data at south and west have returned.  We 
suspect a software problem at the base.

Spent the morning trying to restart communications with central with no luck.
Then also lost communications with west and south and the dsm server stopped
(do not know in what order this occurred).

Could not restart dsm server until communications were restored.  Steve
recycled power on tpop at base, restored communications to all three sites,
and then the dsm server was able to be restarted.

Wireless link to sodar is intermittent.

The PAM mast broke in an apparent high wind event last night; guy wire 
turnbuckles were not tied-off with cable ties!

Lost communication with central.

Repaired mast this morning at ~0930 PST.  Steve restarted etherant communicationby recycling power at central tower.

The data link to central went down around mid-day yesterday; when Steve was
climbing tower??

Went to central this morning & recyled power on both the etherant and the viper.
Link working again.

Network to school down this morning.  Recyled power in attic and back up.

Another TPOP crash, requiring a manual reboot, just occurred.  This time 
everybody came back up (though west took a while).

The TPOP died at 05:15 UTC (9:15PM last night), so there are no local data from
any stations until I rebooted the TPOP just now.

central is down, though the etherant is okay now.  Will have to visit it.

Both south and west recorded data locally through this outage.

P.S. central was working except for the network.  "reboot" brought everything
(including the network) back up.  All data were recorded on the usb, so none
were lost during this network outage. (Also, power stayed up overnight.)

The school connection has been very weak.  I spent an hour on their roof this
morning (in part to help DRI with their 915 Freewave problems).  The best S/N
I could get was 4-6db -- pretty lousy.  Sometimes I can now ping, sometimes not.
Average is ~30% packets get through.

I'm convinced that I need to reconfigure the network to put the TPOP at west.
Guess we'll work on that tomorrow (when Brad is back).

Prior to this, I implemented Gordon's suggestion to lower the RTS/CTS packet 
size from 500 to 32 bytes, but of course this didn't fix school's problem.

lots of network outages now/dropped data.  have changed TPOP to chan 2 and 
rebooted several times. ?????

Two network improvements today:

- Got school connection working (mostly by turning off IPTABLES).  Also found
that this laptop won't reboot due to some docking station power issue.  
Installed (probably very weak) battery to smooth power surges?

- Got south working as we want.  Fixed cable to top etherant, set etherant
to static IP, gateway of AP24 external, removed lower etherant.

Still want to exchange strange cabling at west to AP24 with a good cat5 cable.

This is a stab at defining the network we've settled on:

(see /etc/hosts for complete list of machine names and /etc/dhcpd.conf for
MAC addresses)

.20 subnet (wired; in base trailer; the 8-port hub)
- aster [base computer]
- toughbook
- other users [currently Gordon's and my laptops]
aster is a DHCP server for this network

.12 subnet (wireless; access point is TPOP channel 1 at base)
- TPOP wired interface (also the 4-port hub)
- TPOP external antenna interface(?)
- Prometheus (for hot-films) through TPOP external antenna
- aster (wired to 4-port hub)
- Etherant at Central (wireless)
- nids at Central (wired from Etherant)
- AP24 internal antenna at West (wireless)
- Etherant at Independence Unified School (wired from Etherant)
- "school" laptop in the school attic [our internet gateway]
  (This laptop also on the Independence school local net, but is accessed by
  yet another address using "destination NAT" from Bishop's school.)
- Etherants at lidar sites [not yet installed] (wireless)

.13 subnet (wired interface of AP24 at West)
- AP24
- nids at West

.14 subnet (wireless; access point is AP24 external [parabolic] antenna)
- AP24 external antenna at West
- Etherant at South
- nids at South

The Bulgin connector on South's etherant was damaged during installation, but
we don't want to have to strip it from all the way up the tower.
At the moment, our workaround was to install another etherant at the bottom
of the tower (using rope!), since it still has good line-of-sight to West's
AP24.  This was using another, short, cable.

Brad has rebuilt the pins on the cable to the original etherant, but was 
unable to finish installing the housing and backshell due to intense blowing
dust yesterday afternoon.  On our next visit, he'll finish this task and we
can get back to a normal configuration.  The top etherant will have to be
reconfigured since it is set now to talk to the TPOP (.12 subnet), rather than
the AP24 external antenna (.14 subnet).

We just found that we can only run at 10BaseT speeds using the cabling that we
have to the AP24 at West.  Unfortunately, the AP24 appears to go to 100BaseT
on power-up.  Thus, we have a sequence:
- power-up AP24
- run the "Alico AP24 winbox" application on the Toughbook
- connect 192.168.12.43 (should be default)
- click on "ether1"
- set "10MBPS" and uncheck "Auto..." and "Full..." on "Ethernet" tab
- click on "Apply"
- ssh in from an outside machine (not necessarily easy)
- "reboot" nids

Since this is an unwieldy procedure, we need to either
- make a new cable (cat5 or cat6) that is long enough (18m)
- figure out how to save 10BaseT as an AP24 power-up default

Independence: Joel Hampton superintendent 760-878-2405
Bishop: Joe Griego IT director 760-920-1800 (c) 760-872-7027 (office)
        Justin Norcross Network manager 760-920-8959 (c)