• NodeHollywood

Node name: Hollywood Library
Live Date: June 3, 2006
Sponsor: Lisa Wylie
Contact: Russell Senior <russell@personaltelco.net>

2134 NE 41st Avenue
Portland, OR 97212
Map: NodeMap:NodeHollywood

Equipment

• Metrix Mark I, 9 dBi omni w/ 7 degree downtilt
• Nucab

Software Installed

• Metrix Pyramid
• Standard Builder install on nucab

Network Configuration

• ESSID: www.personaltelco.net/hollywood
• Channel: 11 (2.462 GHz)
• BSSID: 00:0B:6B:4D:77:96
• Nucab Hostname: ramona (10.11.62.1)
• Metrix Hostname: henry (10.11.62.2)
• Internet Network: 68.165.229.145/24
• Gateway: 68.165.229.1/24
• PTPnet Network: 10.11.62.0/25

Installers / Organizers

• RussellSenior

• CalebPhillips

• AlexisTurner

• DonPark

Maintenance and System Log

• 2008-04-18: Noticed today that the wireless had been out since about mid-February. Was able to log into the metrix and found some driver and/or hardware problem: "wifi0: stuck beacon; resetting (bmiss count 4); wifi0: ath_reset: unable to reset hardware: (HAL status 0)". Rebooted the metrix and it seems better. I noticed that my recent stumbling hadn't seen beacons from the radio, which led me to check on it. --RussellSenior

• 2007-08-24: Stopped by today to check on an outage. Wireless has been absent recently, but the call this time was due to a wired outage on a third NIC. Turns out, Keegan's recently upgrades stomped all over the local nocatauth initialize.fw tweaks and they had to be dredged up again and re-implemented. The tweaks consist of the addition of this at the end of /usr/share/nocatauth/gateway/bin/initialize.fw:

  # NAT for local wired network (cribbed from liberace, circa 2006-06-08)
  iptables -t nat -A POSTROUTING -o eth0 -j MASQUERADE
  iptables -t filter -I FORWARD -s 192.168.0.0/24 -j ACCEPT
  iptables -t filter -I FORWARD -d 192.168.0.0/24 -m state --state RELATED,ESTABLISHED -j ACCEPT

  NodeFreshPot has a similar need, as well as probably any other nucab with a third NIC. Coincidentally, the rooftop metrix seemed to have reappeared in the process. I have seen some odd metrix rebooting behavior at another node of similar vintage, they might do with a newer version of pyramid. --RussellSenior

• 2007-02-07: Since around the first of the year, this node has been having trouble. It looks like there was a problem with maybe a flaky crimp. I had not been able to contact the roof, and yet it was clearly on (beacons present, could associate). I was initially not able to connect over the ethernet run to the roof, but replugged into the nucab and it connected. Have to watch it and see if we see more problems. I'm using it right now from the basement. --RussellSenior

• 2006-08-28: Been having continuing problems with routing failures, which seemed to be related to using static rather than DHCP network configuration on the DSL side. The thing preventing us from using DHCP was the provision of the wrong default gateway address. Today I figured out how to override that setting in /etc/dhclient.conf. Switched to DHCP from static, and we'll see how that does. --RussellSenior

• 2006-08-16: Reconnected Kate McPherron down the street whose WET11 had been out since the old node left. The WET11 needed repointing at the new SSID --RussellSenior

• 2006-06-07: Ran new cat5 from the demarc to the DSL modem in the basement. Fixed up lan access with iptables hackery. --RussellSenior

• 2006-06-05: Apparently flakey phone wire to the basement jack. Working on moving to better jack or rewiring basement run. Temporarily dysfunctional. --RussellSenior

• 2006-06-03: Installed today. Metrix is bridging to a nucab, ramona. DSL is provided by Stephouse via Covad. DSL modem is in PPPoE, a Broadxent Briteport DSL modem serves (what I assume is) relay'd DHCP. For some reason, the nucab is coming up without a default route (or it gets lost almost right away), which seems to confuse NoCatAuth. Adding the default route manually and restarting NoCat seems to fix it, but it isn't clear why it is failing to begin with. Also, upload speed (as reported by the DSL modem) is only 96 kbps. That kind of sucks. Also, we need to wire up and configure local access. And bill the node host for the remaining items on the parts list. --RussellSenior

ToDo