FANDOM


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

{{{ All my nodes running dev88 which I believe quite stable release at the moment.

I am using both minsig and blocknode together though latter may change sometimes and I do blocknode again manually. But this is rare since my links stable in general due to short distances between nodes.

We are checking possibles causes for frozen boxes. We identified two causes so far: First is PCI risers used in new MeshBoxes of Locustworld, ie T3. These risers have quality problems on soldering. 20 we have purchased and so far 5 ise faulty. Second cause is memory problems. In two of the boxes we have seen that 128Mb RAMs report different values. One is 14MB, the other is 54MB. First is originally came with LW box. The other we have purchased here. But to my surprise, the box with 14 MB worked nice in general but frozen sometimes. The latter gave segmentation errors during the operation. So we have seen that even with 32MB RAM MeshAP would perform well.

There may be other causes sure, but these two identified as hardware related problems. We are planning to log all activities of boxes to further identify the other causes.

The other problem we have faced sometime was the boxes returning to default settings. We have overcome this by changing "wiana.settings.default" file. Essid, channel and tunneltype set as our network values so when a box return to default it still finds the mesh and downloads new settings from wiana and keep working normally.

We do not use any precaution against power related possible problems yet and I anticipate some problems may be power related. I will add extra capacitors to 12V line soon. Additionally we are preparing PoE devices of our own which will supply 24-48V to boxes and SMPS inside the box will convert and regulate this to 12 volts. On 24-48 side we will put small batteries also in case power fails.

Although we use term "frozen" but in general these are kernel panic situations we guess. Sometime we go with monitors to see the cause and it is kernel panic. But there might be real frozen, unresponsive boxes too. We are not sure. I am looking for a small TFT monitor to carry with.

since I am 15 km away from the network, when a node, especially gateway node stop responding and meshing, it is really a headache to reach and reset them. We are looking for some remote resetting solutions such as GSM controlled relays for at least gateway nodes.

My longest running node is up 25 days but it has a low usage and user. Since we are changing some settings sometime we reboot them as necessary. Once I was using soft booting to see how long node is up without a problem but I thought this might cause memory problems then we have changes again to normal reboots.

Further to the original problem. We are having similar problem with dev 88. I have not really identified the problem of frozen boxes and Kernel panic however it seems in the last two weeks I have been able to minimize these problems. To resolve the freezing problems I had to prevent a number of nodes from switching between one route and another route. What I mean by this is due to reflection and poor signal quality I have nodes that try to by pass other nodes and skip to the uplink node or Gateway node. The MeshAP kept finding a new route to the gateway and broadcast the new route every few minutes. I tried to use MINSIG but since the leaves are falling off the trees and it has become colder I have better signal quality between nodes! I can get a signal to be reflect as strong as a normal signal but the quality may be poor. As a result MINSIG doesn't work in my situation; however, there is more than one way to skin a cat.

I have been using the feature "broadunblock" for nodes to stop blocking signals and found that this was basically the root of the problem. Since multiple nodes would change path every time a tree swayed in the air (giving a stronger signal and as a result a new path way.) What I did was to stop issuing a "broadunblock" command. The nodes that contain poor quality connection, I forced them to connect to the closes node by issuing a blocknode xxx.xxx.xxx.xxx at the Meshap that was receiving the signals. In other words A is the Gateway MeshAP and C is the Repeater that tried to communicate poorly with A. B is between A and C. Now by issuing a blocknode command at A and C thus A talks with B which talks with C. This start to give me some network stability.

I also notice that B started to block C and sometimes gave me a problem. I issued a Broadunblock in Wiana, with B, every 4 hour and this seems to insure that A talks with B and C. The fixed route has stabilized the network. As a result, no frozen boxes or Kernel panic happened during that time. I also had a similar situation with boxes A, D, and E. Once issuing the blocknodes at A and E and the broadunblock at D this too improve the network stability.

Why I was writing to you is that I haven't been able to determine my longest running node as I had to update the notes recently due to a change of IP address in my walled garden.

Are you still having problems with frozen boxes and kernel panics? and what have you done to try to minimize the problem?

> Few times I hooked a monitor and keyboard to frozen boxes and > seen that it > gives a Kernel Panic. But I do not know what causes for this. It is > diifucult to hook up monitor every time to nodes since they are in attics > mostly. > > Some nodes are working without problems. Currently longest run > node is up 21 > days, in past I had nodes live more than 30 days. Some of them we have > rebooted due to maintenance or update purposes. > > Some nodes freezes with no response: No ping, no connection, > nothing. Some > responses to ping requests, works as a mid hop for other nodes > but I cannot > ssh thru mesh or I cannot connect with wireless client. Both do not > remotemanagement. > > I will increase the RAM sizes of problematic nodes and see what > will happen.

> Thanks for the update. I did not see any MeshAP reset them selves to > default settings; however, I am having the exact same problem with the > freezing of the meshap. > > First the MeshAP that are freezing are my compaq hardware not the > VIA epia > boards. Are you running similar hardware or is everything freezing? > > I have instituted a 24 hour period to reboot all the MeshAP starting from > the furtherest hop to the gateway. Currently I only have 12 > meshAP's with 5 > in service. The reason is that I was having a large problem getting the > network stable. > > I have also setup CMUPS from feeed.com So far I have not promoted > my service > because I cannot keep the network running for very long. > > This is what I have done. > > Sent a letter to Jon asking him for a reboot function in Wiana based on > time. > I am rebooting my nodes manually every 24 hour period. The > reason for the > time frame is that I can keep MeshAP alive for 48 hours, after > that time the > problems start to occur. > Next uptime. I need to keep my network up 23 hours and 57 > minutes a day. 3 > minutes slated for reboot. Hopefully this would be acceptable to my > customers! > > If you come across anything or a partial solution please share > this with me?


> > Just now, one of my nodes in that state again: I can ping it, it > > lets other > > nodes connect to gateway over itself, but neither I can ssh nor > > any client > > connect wirelessly. It does not do remotemanagement. It stays > > same until a > > hard reboot. > > > > We do not know exact reason yet, even we are not sure if all > > frozen events > > are from the same cause, or hw/sw related. We just anticipate > > that RAM would > > not be enough. We use standart boxes with 128MB RAMs. Some of them 32MB, > > some 64MB CFs. > > > > LW cannot offer any solution/diagnostic for this problem. I > > suspect memory > > management problems after hours/days of operation. Probably I > > will test them > > rebooting once in a day to see what would happen. > > > > Actually there is one more major problem in addition to frozen > > boxes: Boxes > > returning to default. It is solutionless again. But this is rare.

> > Yeah reread your problem and I have the same problem and > > environment. Did > > you find a solution to the frozen boxes?

> > > I am running 20 node mesh, all runnning dev88 except one ehich > > > runs on dev > > > 85, no major problem as yours. > > > > > > the biggest problem we face up to day is frozen boxes which > still we do > > > not know the exact reason. Node is alive, you see the signal with > > > sigspy, > > > even sometime you can ping to it but neither customers can > > > connect nor we > > > can ssh into to reboot it. > > > > > > Now we have over 300 hundred users authenticated by their MAC > addresses > > > only. We do not use VPN or WEP, only clear connection. > > > > > > We guess our problems cause from the RAMs we use. They 128MB, and when > > > there are heavy traffic and user activity, memory is not enough during > > > swap/unzip or whatever it does, we think. But this is the > > > assumption only, > > > no certain evidence. > > > > > > second major problem we have faced again, though not often as > first, is > > > boxes that return to default settings. When it defaulted, essid > > > changes to > > > Locustworld at channel 1, lost mesh connection and we cannot reach it > > > remotely. So we have to climb to reach it physically. We have changed > > > default settings file so it gets our essid and channel in > such cases but > > > still no solution to this. > > > > > > As you see system is in an acceptable reliability even though > there are > > > problems that is unacceptable in a paid environement. I > recommend you to > > > get professional support from one of the experienced guys near to you. > > > Since you run on bank credit, you have to be very carefull in your > > > operation, I think. > > > > > > Hope you find a quick fix to your problems. In US, Tom Sharples > > > is one of > > > the experts that might off help to you. > > > > > > Let me know if I could be off help.

> > > > Sorry to hear of your problems. > > > > So that we might avoid same, can you let us know what > hardware are you > > > > using ? > > > > > > > > We took the decision early on to stick with proprietary off > the shelf > > > > equipment in an attempt to minimise the variables i.e. the > LocustWorld > > > > Mesh Box and Darklite and Netgear Client equipment. > > > > > > > > These have been out in the field for over a year in attics, > Barns and > > > > Loose boxes running build 25 dev 88 and now 90 > > > > > > > > Phil > > > > Digital Parish > > > > > > > > > > > > This Email has been checked by Norton Symantec AV > > > > This email and any attached files are confidential and copyright > > > > protected. > > > > If you are not the addressee, any dissemination of this > > communication is > > > > strictly prohibited. Unless otherwise expressly agreed in writing, > > > > nothing > > > > stated in this communication shall be legally binding.

> > > > Food for thought to any new wanta be WISPs, > > > > well just as a note for what ever it might be worth... i have been > > > > running dev76 for about 4 months and tonite after fighting > > > login issues > > > > for 4 months (put a a customer on and lose a customer because he > > > > couldnt > > > > log in)... i finally decided to try to upgrade to dev85 > > > since a friend > > > > of mine is running that on his mesh which has quite a few pcengine > > > > boards on it. i picked the least used node on my mesh.. ran > > > > getandverify tobuild25dev85... had to get the patch.. got it.. ran > > > > getandverify tobuild25dev85 the damn thing downloaded little > > > over 4 meg > > > > .. i lost my connection in putty(signal level at 104 in > wiana).. so > > > > what > > > > ever happened over the next hour i have no idea.. i drove 14 miles > > > > round > > > > trip to power it down and back up.. its DEAD as a hammer and > > > 156 feet > > > > in > > > > the air... im stuck with what ive got but since its free > > > software what > > > > can i say.. you get exactly what you pay for! its just a > shame all > > > > that > > > > stuff is tied into a T1 line that costs 750 bucks a month > and a year > > > > contract which is the only thing that works the way its > > suppose to(you > > > > get what you pay for)... > > > > this is a kewl toy but i dont think its even close to a > > good beginner > > > > business solution... i started out with a wap11 as an > > encrypted ap and > > > > 7 wap11 clients tied straight into the T1 line and a good > > > reputation > > > > in > > > > my community 6 months ago... now i got 30,000(lost 5000 in a one > > > > lightning strike) dollars tied up in infrastructure towers > > UPS's mini > > > > itx's and pcengine boards and still 7 clients and they get less > > > > bandwidth and cant log on half the time and when i try to sell a > > > > customer service i can tell my reputation aint so shiny > > > anymore.. all i > > > > wanted was something that would JUST WORK so i could make a little > > > > money > > > > and buy some really good stuff later > > > > dont worry i aint giving up on the mesh... i have no other > > options at > > > > this point but i think a little constructive cristism was in order > > > > since > > > > the bank that loaned me the money to build this whole joke > > > bought their > > > > wireless service from another company that just put in a > regular ole > > > > wireless AP last week because mine didnt work.. YOU GOTTA laugh > > > > because > > > > my bank is paying 25 dollars more a month than what i charge.. > > > > btw this in no way voids any of the help you guys have > > > offered me.. i }}}

Ad blocker interference detected!


Wikia is a free-to-use site that makes money from advertising. We have a modified experience for viewers using ad blockers

Wikia is not accessible if you’ve made further modifications. Remove the custom ad blocker rule(s) and the page will load as expected.