WrapAndPoe

back to http://scratchpad.wikia.com/wiki/Sasecurity

MAKE your OWN POE
1. http://www.nycwireless.net/poe/ 2. http://www.gweep.net/~sfoskett/tech/poecalc.html 3. http://kawasaki.kz/wireless/PoE.htm

WRAP timing issues no onboard clock
>Check to see that all of your nodes are on the same version of firmware.

I agree!

>It was theorized by some on this list that since the WRAP lacked a proper onboard clock that some timing issues might arise and it was important to verify that the date and time were correct on that node, as well as all of the others.

The Wrap lacks battery for keeping the RTC running between reboots, however this is not a problem since the time is synced at each startup. But the standard "open source" software's way of setting the time after syncing with wiana, sometimes causes a lock-up of some boards. We have addressed that (and other issues) in our patched version.

I've seen this frequently when I was using the open source software. Check to see that all of your nodes are on the same version of firmware.

This problem occasionally cropped up still on the 3rd hop after I moved everything to dev76. Everything on my network was running on VIA EPIA motherboards except the first hop, which was a PCEngines WRAP board. I removed that unit from my mesh and I have yet to see the problem again. It was theorized by some on this list that since the WRAP lacked a proper onboard clock that some timing issues might arise and it was important to verify that the date and time were correct on that node, as well as all of the others.

So, check to see that all firmware is uniform; check to see that no hardware issues are causing this; and make sure your date and time is correct on all nodes (even though Wiana should update them automatically).

> I don't understand this, and it's a really annoying problem. > > I've got a number of meshboxes with this problem intermittently, but > one that's particularly bad right now. This meshbox has a route to the > gateway machine, but won't establish a tunnel with it, and so can't > communicate with the Internet. The gateway is listed in "reporter" as > being three hops away, which is correct. I can ping the gateway's > wiana address, 1.190.108.93 with no packet loss. Every minute I get a > broadcast message "Attempting to connect gateway 1.190.108.93". > > TUNNELTYPE is IP in wiana.settings. > > What's the problem?

asdfasf
I totally agree with you that the mini itx is a VERY much better solution than the wrap board. All i was addressing was practical power issues and throwed the pcengine wrap board example in to possibly solve some others problems with the 12 volt power filter problem i had discovered and fixed while using POE. Personally any mini itx will have either commercial power at its mount point or i will run #8 wire to power it off of car batteries and a battery charger since i kill the UPS backup problem and some lightining issues since i can apply the "faraday sheild effect with batteries" at the same time. Its my opinion that half way "half engineered" and "may get by installs" will fail. By the time a person reengineers designs changes and experiments and suffers several mysterious failures the price of the #8 wire which is a "put it in and forget it forever solution" is worth the effort and the small cost. But the closest node i have to the ground is 128 feet and the tallest is 158 feet and they all communicate at above 100 on wiana scale(whatever it is) and one node is 12 miles away from the farthest one it talks to and i have 7 in the air now which all are pcengine wrap boards i put together myself except 1. i do use a mini itx for a gateway but its on the ground and transmits up the tower to the first node to isolate my T1 router and servers from the tower due to equiptment loss because of a direct lighning strike i took. I do plan to upgrade to mini itx's but a poor man has got to do it a poor mans way. And since i only have a limited amount of money the wrap boards are all i could afford and have served me well and became EXCEPTIONALY STABLE once the power was filtered with a 4700 ufd capacitor(priced at less than 4 dollars US and an additional 10 minutes to install it). btw if anybody has some low class pcengine wrap boards they dont want give em to me i will be happy to pay the postage! i guess i could go FIRST CLASS but i just couldnt go as far and stay as long as the rest of yall. Im kinda like Johnny Cash im buidin this thing ONE PIECE AT A TIME! (BTW Tom let me know how the 200 foot version of your poe solution works out! My personal email address is my back ground is in RF and analog power supply design so it might be interesting to talk shop and swap notes if ya like) Are you having any problems with induced spikes into the ethernet cable when the multivibrators are switching.. this was the problem i was encountering in the prototype i built and was testing at 100 feet. I did try ferrite beads but the problem seemed to be the varying load along the entire length of the ethernet cable and its inherent proximity. If you do have this problem let me know and i will offer you a possible simple solution which i never tested but has theoretical merit. I lost interest in the project and decided it wasnt worth the effort for just my own personal use. Oh i also looked at a 180 volt dc to 12 volt dc 60 watt converter to get the current low enough for the ethernet cable to be more effiecient but thats a whole other story since in the back of my mind i kept thinking about condensation at the power supply ethernet plug and electrocutin a computer geek by accident since they dont know anything about anything over 24 volts lol

=
WOW what brand ethernet cable are you using Tom? How much voltage drop do you get per foot with your ethernet cable?? How long are the runs? I use poe with pcengine boards. And even then i have to filter the Power if the run is over 120 feet with at least a 4700 mfd capacitor to keep the spikes out of the power when the radio transmits otherwise i get some REAL SERIOUS packet loss LOL.. but to see is to believe

=
=   How long was the ethernet cable?

>Interesting problem with a MeshAP radio consisting of a WRAP 2B board and the WRAP POE injector. We purchased this radio from a "vendor"..we didn't build it. > >We plugged the radio directly into a power supply and everything worked fine. > >We then mounted the radio on the roof using POE and we could "see" the radio. But we could no get past the internal network (router). > >After scratching our heads and trying everything imaginable for 3 days, we brought the radio back down and plugged it into the wall and everything worked fine. > >We then switched the POE that came with the radio with an old POE (generic) and everythiung worked fine. We then tried a Smartbridges POE and again the radio worked flawlessly. Yes, the POE was the last thing I would think of in a gateway problem since the radio could transfer internal network data fine. > >I've never seen a radio actually work but that could not get past the gateway due to a POE (?). However, switching the POE made the radio see the gateway and allow users to surf the net. Switching back to the WRAP POE and all the users could reach was the router. Yes we tried different wiring patterns in our cable but to no avail. > >Lesson learned for us, we have nixed the use of WRAP POE injectors.

=
==================================== I also had this problem. The issue as I saw it was that my wrap POE shipped with a 9volt Power adapter. I swapped it for a 12 volt and all was fine. I think with dual semao 200mw cards and 200 foot of cable it was to much for the 9 volt adapter. The other thing a friend saw recently was that he received a center negative adapter that did not work either.

sdfgasfasfasfas
Regarding the power over ethernet, I tried many power supply (15v and 18v) plugged in a stabilized battery powered UPC.

>Are you running this node with power over ethernet? If so, try changing the >power to a higher rated power supply or try connecting the power to the unit >directly and see if that cures the problem. > >If you're not running the latest version, you can get it by logging in to the >node and issuing: > >getandverify tobuild25dev106

>>Why does this happen? I do have a few other wrap repeaters with no reboot >>problems just the gateway. I haved swapped the gateway with a new one with >>same problem and it is grounded. Any help with this is needed. TJ

I have the SAME EXACT problem, mainly the gateway (because of the load I assume) although it occured on a repeater as well... On the bright side, it restarts automatically...

I already ruled out the following:

- temperature (it happens at different times of day, hot and cold) - high traffic (it happens sometimes with very little traffic)

I suspect:

- A watchdog that gets triggered (although I disabled the internal watchdog and it didn't fix the problem) - A weird traffic, like some ICMP or UDP or DNS traffic or other that confuses the gateway

> Why does this happen? I do have a few other wrap repeaters with no reboot problems just the gateway. I haved swapped the gateway with a new one with same problem and it is grounded. Any help with this is needed.

Why does this happen? I do have a few other wrap repeaters with no reboot problems just the gateway. I haved swapped the gateway with a new one with same problem and it is grounded. Any help with this is needed. TJ

=
The node will loose settings, but once you reflash, it should get the same hardware key (I *think*) when it first checks back in with Wiana. At least that's been my experience. The hardwarekey is based upon something with the CPU and ethernet port (there again - I think!).

Anyway, I've reflashed nodes and got the same IP back, downloaded the old settings from Wiana, and been back in business.

> > I did the getandverify tobuild25dev106, saw it download and the node > say it was going down to a reboot, but it never came back. > > Have stuck a known working CF card from another node in and that works, > so assume the hardware is ok. > > Anything I can do / test? > > If I cut my losses and flash the CF card manually, will I loose all my > settings (eg had hard coded a fixed IP address for the Ethernet > interface), or will they get pulled down from Wiana?

> Are you running this node with power over ethernet? If so, try changing > the > power to a higher rated power supply or try connecting the power to the

All my power supplies are 12volt 165amp Gel Cell batteries withless than 20 feet of CAT5 POE. And a 6 amp trickle charger for the battery.

I have been getting help from Jon and Dave from ultramesh. It seems my gateway crashes were being caused by me running 'watch reporter', and the funny/sad thing is that the more it 'crashed' the more I 'watched' and the more it crashed.. Ugly. But now it is solved. Thanks to Jon!!

As for my WRAP crashes, I had PCMCIA on and faster bootup off, switched both of those settings, set my unblockretest to disabled, and minsig of 95. Since then I have not crashed. Thanks to Dave!!.

Sometimes I think it just takes a fesh set of eyes to look at the problem. Lets hope this fixes it.

I am wondering the size of peoples mesh.. Perhaps we can all tell big we are.

18 nodes(12 dual and 5 single), 5 gateways(1 big mesh with 3 g/w and 2 little meshes). 115 clients. Started in Oct '03