Jutta emailed a couple of days ago saying that stuff's down. Michael went by, rebooted the gear in the kitchen using the powerbar behind the fridge, and it all seemed to come back fine.
On Nov 5 both wifidog and open-mesh reported that the routers at DGP were down. Gabe contacted the folks at the park via email, and they confirmed that while they were moving the computer, they ran into equipment trouble. The power supply for the wimax modem broke again, and Rogers refused to replace it (instead offering to sell them a new modem for $100), so they got a new power supply (at Active Surplus?). In the process, the wired router stopped working. I went by today to check it out. It looks like the (5V) USR wired router died, perhaps because the (12V) Linksys power supply had been plugged into it. I plugged the modem into the Linksys router, whitelisted the rinkhouse Dell workstation and the open-mesh gateway router in /etc/wifidog.conf, and now things are working. The USR router has been pitched. Gabe will pick up a new wired router sometime soon and come back to plug it in.
The mesh network at DGP went down a couple of days ago. I emailed staff to ask them to powercycle the rinkhouse router, but the network didn't come back up, so I went by to take a look. There was no signal outside the rinkhouse. I powercycled the router, and it didn't come up. I got a ladder to take a look at the router, and the LEDs weren't on. I took the router down and plugged it directly into the powersupply and the lights came on. So I think the problem is one of: 1) The ethernet run from the kitchen (where the power supply is) to the rinkhouse router is damaged. 2) The PoE injector is faulty. 3) The internal PoE splitter inside the router is faulty.
It's almost certainly the Ethernet cable. I didn't have a spare patch cable to test with. I think we should just go back with a patch cable and a length of cable and a crimper – test with the patch cable, and assuming it works just right away run a new Ethernet line. Bring lots of cable clips, and a hammer drill – if the cable is secure in more places, it's less likely to be damaged from strain. I think.
Wifidog and Open-mesh started reporting that the DGP nodes were down again starting a few weeks ago. (The network ran for about a week after the outage which ended July 19.) Eventually the folks at the park identified that the modem was the problem. They called the ISP and had them reassign the service to another router, which also didn't work. Gabe went in a few days ago to test both modems, knowing that the power supplies flake out frequently. Of the two modems and two power supplies they had, one of each worked. But it wasn't the modem that had the service associated with it.
Yesterday I got word that they got the service reassociated with the working modem, and that it was working. The wifi routers were still offline. I went by this morning, and found that the modem was plugged directly into the computer, and both the interior wired and wireless routers were unplugged. I plugged stuff back in the way it's supposed to be, and everything came up. (The mesh network took a while; around 10 minutes per node?)
I also installed a new power supply and PoE injector for the hollow router.
Both the wifidog router and the open-mesh routers all went offline a few days ago. Gabe went by today. The USR wired router had flaked out – the 'power' LED was turning orange, not green. I did a factory reset (held down the reset button), reconfigured the router (set the IP address, admin password and turned wifi off), and everything started working again.
We installed OM1Ps at the rinkhouse, hollow and south end of the park, which gives us good coverage. Currently the modem is plugged into a wired router. The rinkhouse OM1P, interior wifidog router and staff computer are connected to the wired router. The open-mesh network name is wt-dgp. Soon after we installed it, the power supply and PoE injector for 'hollow' went missing. We haven't yet replaced them.
Wifidog reported the node down for >10 days, so I emailed people at the park to ask what was up. They replied saying they've been having problems. I went yesterday and the wimax modem appeared to be busted – when plugged in, all the lights came on dimly, and didn't flash in the way that indicates it's acquiring a signal. I took the modem to the office and plugged it in with a different power supply and it worked. So today I went back to the park with the working power supply and now it's all working. I also took a look at the outdoor router, which is still having the super-low signal strength problem. I opened the box – everything seemed ok, except that there were some bugs inside, and a little corrosion on the boards – some water got in. I tried every possible antenna configuration, with no improvement. Either both antennas are defective, or both antenna connectors on the router are defective. The ideal solution would be to bring a new, waterproof box with new antennas and a new router – replace everything and troubleshoot offsite. A job for the spring, perhaps.
I got an email a few days ago from a user saying that he couldn't find the wifi in the park. I told him where to look for it, and he still couldn't. So I went by to look and, indeed, there's no wifi in the park. The router inside the rinkhouse works great, and the outside one works as long as you're within about 10 feet of it. The farmer's market was going on while I was there, so there was no way I could climb up and look at it. Sounds like what would happen if the antenna was unplugged. I'll go back sometime soon with a new antenna and troubleshoot it.
It's taken me (Gabe) this long to get back to the park. I got the outdoor router running by replacing the defective Ethernet extension cable in the north kitchen. Everything's working now.
Dafydd and Gabe went to the rinkhouse today to run two cables: one from the skate room to the snack bar/kitchen, and one from the snack bar/kitchen to the computer cabinet. We also moved the modem and router from the shelf above the stove to the top of the shelf across from it. The router showed a link to the outdoor box, but the two devices couldn't ping each other. I'm not sure if it's a cable issue or a config issue… further investigation is required. But after spending all morning in 32 degree heat at the top of a ladder, I didn't have the energy to figure it out. The computer works using the wired connection; I whitelisted it in the wifidog.conf
The LAN interface of the outdoor router has the IP 192.168.1.10.
Mayssan emailed Michael & Gabe to say that the wifi has been shaky/intermittent. I (Gabe) went in this morning to check it out…
Solutions that we came up with:
Gabe is planning to go on Tuesday (26th) morning at around 9:15am to install the two cables: a 100' one for the long run, and a ~50' one for the short run.
Gabe removed the equipment from Bruce's house today.
MAJOR UPDATE: the dsl line (like the one at Dundas Square, coincidentally or not) doesn't seem to want to stay up for more than a few hours. We discussed it with park staff, and decided to order the wimax service early. Michael and Gabe went to the park today to set it up.
The wimax modem is in the skate room, connected to the router labelled 36 (but with the wifidog gw_id of 31). We reconfigured that router from “repeater mode” to “normal wifidog mode”. We then did the opposite with the WT router in the exterior box: converting it from a wifidog node to a “repeater node”. No changes to wiring were required.
Gabe disabled the SSID broadcast on the south router, to avoid confusion. He also marked the node 'temporarily closed' on the auth server.
Wifidog reports both routers going down at about 6:30am.
The tentative plan is to switch to wimax in April/May 2007, locating the modem at the rinkhouse. The friends of DGP will be able to cover the cost, using money earned from their snack bar.
South router still down, and on the weekend (26th or 27th?) the north router went down. Patrick, Dave and Gabe went to the park to check it out. Connecting to the rinkhouse router, we could ping everything within the park (except the south WT router – packets don't route that way)… which means the mesh is up, and either it's a routing problem with the south Nomad router, or a problem with the DSL modem or line. We walked over to Bruce's, and were lucky to find him home. He let us in. The WT router wasn't sending out beacons. Gabe plugged his laptop into the dsl modem, and it seemed to be working ok. We powercycled the modem and the routers, plugged it all back in, and everything came right back up. No explanation for what was wrong before. (Gabe's only theory: the DSL modem crapped out and needed a reboot. On Jan 12th when Bruce powercycled, he did the routers only, not the modem. This doesn't explain, though, why the south WT modem wasn't sending out beacons.) We left the park happy.
Bruce Whitaker powercycled the routers at his house. The north router came back online, giving park staff internet access, but the south end is still offline. The mesh link between the two ends is functioning. Access to Bruce's is necessary to figure out why our router there isn't working.
Router 31 was reported as down in wifidog for most of the first week of 2007; Michael P investigated on Friday night, and reported this:
The antenna on the external rinkhouse routers had been knocked out of position and was, presummably, not seeing the south end for its internet feed. A broom handle and power cycle of the routers (plus the one in room 36) helped set it straight.
Oddly, south is reported being down right now though the rinkhouse looks fine.
Indeed, router 30 is reported as down in wifidog, and Gabe couldn't reach it via openvpn – seems like it'll need a powercycle.
DSL and cross-park “mesh” provided by Wireless Nomad. DSL modem owned by WT. Paid up until early/mid-May 2007.