Differences between revisions 9 and 10
Revision 9 as of 2014-03-26 00:56:47
Size: 3473
Comment: add some notes about when openvpn dies
Revision 10 as of 2014-03-26 10:14:28
Size: 3808
Comment: add accounting section, move some information there
Deletions are marked like this. Additions are marked like this.
Line 21: Line 21:
 * Running low on Buffalo WZR600DHP, may be approaching time for another batch of 10, current price on new stock is ~$80. Have some payments still pending: $80 from NodePotatoChampion, $70 from NodeKnow, $150-ish from NodeBluebird, $150-ish from NodeWygant (waiting for roof install to bill).  * Running low on Buffalo WZR600DHP, may be approaching time for another batch of 10, current price on new stock is ~$80.
 * Accounting:
  * Deposited today:
   * $75 from Tom for NodeArborLodgeFund
   * $70 for a Buffalo for NodeKnow
   * $70 for a Buffalo for NodeHancock
  * Invoiced but not yet received:
   * $80 from NodePotatoChampion for Ubiquiti Bullet M2HP
   * two more contributions for NodeArborLodgeFund
  * Not yet invoiced:
   * NodeBluebird, $150 for a Buffalo and a Ubiquiti Loco M2
   * NodeWygant, also about $150 for a Buffalo (installed) and a Ubiquiti Bullet M2HP on their roof (still pending)

Location: NodeLuckyLab
Date and Time: Wednesday, March 26, 2014, 6:30-8:00 p.m.
Scribe: You
Roll Call: Them


  • Server transition report
  • Node notes:
    • NodeHancock -- new node plugged in

    • NodeWygant -- still needs a roof plugged in

    • NodeSunnyside -- needs a mast, either on a fence post or a gable-end (waiting for a photograph to assess viability of the latter)

    • NodeHawthorne -- plugged back in after an extended outage due to electrical code problem

    • NodeBluebird -- needs some help from the node host, to remove a malfunctioning upstream router

    • NodeArborLodge -- two of three (at last po box check) contributors slow to contribute negotiable currency or instruments. Need to ping them again with ultimatum. No payment by end of billing cycle means we're turning it off, refunding pre-payments.

  • OpenVPN-related node outage early on Monday evening, debugging a problem we are seeing with ipv6 olsrd, where something seems to cause routing table collapse. Sometimes, turning off olsr6d on specific routers seems to cure the problem. Reported upstream, with strace and packet capture. Unclear what is causing it. Restarting openvpn (needed to avoid conflicting vpn addrs) on iris can/does result in carnage across PTPnet. Also, may want to implement --ifconfig-pool-persist on iris openvpn, which will require another restart.
    • Nodes show this in their logs:
      • Mar 24 16:22:35 ${ipaddr} openvpn(ptp)[830]: Linux ip addr del failed: external program exited with error status: 1
      • Mar 24 16:22:37 ${ipaddr} openvpn(ptp)[830]: Note: Cannot open TUN/TAP dev /dev/net/tun: Permission denied (errno=13)
      • Mar 24 16:22:37 ${ipaddr} openvpn(ptp)[830]: Linux ifconfig failed: external program exited with error status: 1
      This is because the process has dropped priveledges and is being asked to reconfigure an ipaddr. Apparently --ifconfig-pool-persist is the suggested answer.
  • Running low on Buffalo WZR600DHP, may be approaching time for another batch of 10, current price on new stock is ~$80.
  • Accounting:
  • Nodes with older images have damaged olsrd start/stop logic, need to investigate/mitigate or pave over.
  • Russell experimenting with UDP-based dyndns system, involving socat on the server, ncat on the nodes, uses a shared-secret to validate reports, need dyndns on public ipv4, ipv6 and vpn addrs.
  • Lots of new patches in ptp-openwrt-files (aka files-master) git tree of late, review.
  • PortlandMeshNet planning a meetup in St Johns this Sunday, March 30th, 12:00 at McMenamin's Pub in St John's (8203 N. Ivanhoe) https://groups.google.com/forum/#!topic/portland-meshnet/GaAz9luMoRo

  • New splash rollout planning
    • One node so far (NodeLuckyLabNorth), another NodeHancock imminent

    • Bugs/features:
      • Node information link is wrong (needs a leading Node$Nodename)
      • Need better behavior with android splash navigator, needs a fake original destination somehow
      • Maybe extended the timeout from 900 seconds (15 minutes) to something longer to accomodate sleeping devices not being visible to node.
    • Need to contact node hosts with heads-up about impending deployment
    • Start deploying


  • None Yet


MeetingMarch2014 (last edited 2014-03-26 16:15:19 by RussellSenior)