Too much downtime involved in server restart.

[kfogel]

Certain maintenance operations involve a server restart (e.g., the temporary enablement of the /signup route for account creation, which of course will go away once issue #138 is resolved). Those server restarts involve

$ ./node_modules/.bin/forever stopall
$ ./node_modules/.bin/forever -da start --watchDirectory . -l forever.log -o out.log -e err.log ./bin/www

...and the second step often takes too long to bring the server back up and/or needs to be run twice. There is a possibility that the real issue is that the first step takes too long to complete (i.e., the prompt comes back, but the stopall command is not yet actually done stopping everything), and therefore the first time the second command is run it's a no-op because the stopping is still going on. This needs some more investigation.

But basically, restarts should be near-instantaneous. Whatever we need to do to accomplish that, we should do.

[kfogel]

Yes, further experimentation seems to confirm the hypothesis above: wait a little longer after stopall, and be rewarded with a much faster startup time in the start command.