Location: Jitsi
Date and Time: Wednesday, January 8, 2025, 5:30pm-8:00pm
Scribe: You
Roll Call: Them

Agenda

• Pick location for next weekly
  • Encountered someone who wants to get involved with Personal Telco, who lives in N Portland, so some place convenient for them is good for an indoctrination session.
• Node notes:

  • NodeEleven -- there was an outage in mid-December, by the time Russell actually arrived to check, the upstream internet gateway had been rebooted and that fixed the outage. They use a Verizon Wireless 5G home gateway thingie.

  • NodeKlickitat -- Russell has noticed that some of the Archer C7-based mesh nodes have a habit of becoming unreachable after several days. Nothing obvious in dmesg, can usually connect locally and reboot, which fixes mesh link failure. Luckily, there isn't a lot of active use of these devices at the present time.

  • NodeEdie -- still needs Russell to configure a device to install, slacker!

• OpenWrt shenanigans:

  • Russell has added a script (/usr/bin/wiki-info.sh) to PTP devices that generates wiki-pastable network configuration data. He updated a bunch of node pages December 16-17, updating outdated information that had been lingering.

  • Over the holidays, Russell managed to hunt down the bug that was causing the Soekris net4826 (SC1100-based) boards to hang during a warm boot, requiring a power cycle to recover. After the recently decommissioned net4826 at NodeFirstUnitarian, there is only one actively used net4826 in our network: the one at MississippiWest. It gives us some trouble when ssh'ing because of deprecated protocols, so it would be nice to update, and now we can.

  • Russell has been working on the reverse engineering the Realtek-based switches PoE control protocol. Currently has a Zyxel GS1900-10HP v2 splay out on a desk with a logic analyzer attached, to try to make heads and/or tails of the protocol. It is different, although similar to a protocol used to control Broadcom PSE (Power Sourcing Equipment) chips on version 1 boards. The version 2 boards use a different PSE, manufactured by Realtek themselves. The same protocol almost certainly applies to the version 2 Engenius EWS2910P. Some preliminary work occurred two years ago, trying to poke up those involved to finish off the support.

  • Russell gave a PLUG talk on January 2 covering the flashing of the Linksys E8450. In the process encountered a bug in an early production version that had trouble warm booting. Eventually discovered in a 6000+ post thread on the OpenWrt forum, that it was a known problem with a relatively simple fix, setting a minimum CPU speed of 600MHz, which apparently keeps the particularly sensitive memory adequately powered during the warm boot. Later production versions don't have the problem.

• Your Item Here

Notes

• None Yet

[CategoryMeetingNotes]