So Cal Free Net.org

Our intrepid band of volunteers really pulled through today with the most complex single-day install we've ever accomplished!

For starters, there was the common room that become a computer lab. Then there was the dual node installation of a main AP and a second rooftop repeater AP. These were all driven by a newly installed router, firewall and captive portal box (a soekris box running m0n0wall). Many, many thanks to all of you, both the regulars and the newcomers. With one less person, it would have been hard to finish in one day.

Things that went well

We've learnt a lot in the past year and are applying it more consistently. In particular, we've found that the more work we can do before getting onsite, the smoother the day will go. This allows more time to cope with the unexpected.

Some specific things we've learnt previously and applied well this time:

• a solid site survey with the intended installation equipment, including antennas. This meant we didn't have to do any testing or relocation of gear on the day.
• timely purchase and assembly of all the equipment -- physically assembling all the gear beforehand eliminates surprises like: missing small but critical components (zip ties, sealant goop) and catches problems like incompatible mounting hardware (though see below)
  
• network planning -- we designed and fully documented the network layout, including all equipment interface IP assignment, subnets etc. We've always done the design before, but never documented it so well (still room to improve)
• preconfigure settings and end-to-end test, and document it -- this was probably the first hard learnt lesson and we've taken it to heart, though this was the first time we'd documented it. Each piece of gear was configured according to the network plan and documented as we went (gateway router, HS3000 AP). Then, and this is the crucial step, it was all hooked together and connected to a live internet feed. However even this wasn't a perfect test, see below.
  
• project planning -- Also for the first time we put together a planning document. This was the largest install by far that we've ever attempted in one day. We usually do one node, period. This was two nodes and a lab. So we decided to split out the tasks and divvy up responsibilities ahead of time. This allowed some of the work to be frontloaded before the day and also allowed tasks to proceed in parallel on the day. We have more work to do on this approach, but it was a promising start.
• start early -- 10am was on the early side for an install day, but with the ambitious schedule we planned, it was a good move.
  
• standard gear -- for a while we've tried to find a standard set of gear and stick to it. This simplifies ordering and stocking (e.g. we always use 48" LMR 400 N-male to N-male cables and hence N-connectors on all radios). Although we used new radios this time and a new relay approach (WDS), the rest of the install followed a familiar pattern: SBC m0n0wall based gateway/router, PoE powered rooftop gear, lightning protector and 8 gauge copper ground to a solid ground rod or equivalent, our usual Superpass omni downtilt antenna and small stuff like UV-rated zip ties, hose clamps for mounting, duct goop for waterproofing etc.
• volunteers rule -- Last but not least, our many skilled volunteers made the day the great success it was. Given how we're structured and our educational emphasis, we often don't know until the day exactly who will turn up apart from the core team. However it never fails (so far!) that everyone who does show up brings some vital piece of gear or some special skill that makes a unique contribution to the day.

Many, many thanks to the volunteers who showed up on the day and who provided critical support beforehand, including: Seren, Drew, Jason, Michael, Dave, Dlan, Stewart, Marc, Joe, Ben, Matt and Mike S. Not to forget our president Lee, who also wrote a wonderful thank you. All your contributions were vital.

Things we learnt

Although all's well that ended well, there were some 'learning opportunities' during the day. The reason for mentioning them here is not to dwell on them, but rather to make note so we can avoid these problems in the future if possible.

• always put all the gear together before getting on site - in this case the 1" mounting poles didn't fit securely in the 1" mounting brackets leading to a last minute scramble to find an appropriate mounting solution. Fortunately we've learnt this lesson already with the electronics, and all of that worked great, with no onsite configuration necessary.
  
• Double check the configuration as applied - for the first time we completely documened the planned network IP layout, but the gateway WLAN IP was erroneously set to the base network address of 10.12.11.128 instead of the first IP value of 10.12.11.129. Strangely, this worked during testing and also for the laptops we tried which were running WinXP, but not for the lab computers rnning Win2k. (The symptom was DNS not working).
• Plan on making an extra trip to Home Depot - the best laid plans oft go awry. Ideally a trip offsite to get odds and ends won't be necessary, but better to plan for one anyway and then be thankful if you don't need to go.

Equipment Used

a work in progress - the final tally is still being worked out