A privacy-aware, distributed, open source social network.
AGPL-3.0 License
Bot releases are visible (Hide)
Published by jhass about 11 years ago
Published by jhass about 11 years ago
Published by jhass about 11 years ago
/a/b/c/config/boot.rb/../../Gemfile
to require and open things, create a proper path instead.public/default.html
#3811
server.db
to server.database
to match the example configuration.debugger
since it was causing bundle problems, and is not necessary given 1.9.3 has a built-in debugger.Published by jhass about 11 years ago
Published by jhass about 11 years ago
Published by jhass about 11 years ago
Published by jhass about 11 years ago
Published by jhass about 11 years ago
We replaced our queue system with Sidekiq. You might know that Resque needs Redis.
Sidekiq does too, so don't remove it, it's still required. Sidekiq uses a threaded
model so you'll need far less processes than with Resque to do the same amount
of work.
To update do the following:
Before updating (even before the git pull
!) stop your application
server (Unicorn by default, started through Foreman).
In case you did already run git pull
checkout v0.0.3.4:
git fetch origin
git checkout v0.0.3.4
bundle
Start Resque web (you'll need temporary access to port 5678, check
your Firewall if needed!):
bundle exec resque-web
In case you need it you can adjust the port with the -p
flag.
One last time, start a Resque worker:
RAILS_ENV=production QUEUE=* bundle exec rake resque:work
Visit Resque web via http://your_host:5678, wait until all queues but the
failed one are empty (show 0 jobs).
Kill the Resque worker by hitting Ctrl+C. Kill Resque web with:
bundle exec resque-web -k
Don't forget to close the port on the Firewall again, if you had to open it.
In case you needed to do step 2., run:
git checkout master
bundle
Proceed with the update as normal (migrate database, precompile assets).
Before starting Diaspora again ensure that you reviewed the new
environment.sidekiq
section in config/diaspora.yml.example
and,
if wanted, transfered it to your config/diaspora.yml
and made any
needed changes. In particular increase the environment.sidekiq.concurrency
setting on any medium sized pod. If you do change that value, edit
your config/database.yml
and add a matching pool: n
to your database
configuration. n should be equal or higher than the amount of
threads per Sidekiq worker. This sets how many concurrent
connections to the database ActiveRecord allows.
If you aren't using script/server
but for example passenger, you no
longer need to start a Resque worker, but a Sidekiq worker now. The
command for that is:
bundle exec sidekiq
The only gotcha for Heroku single gear setups is that the setting name
to spawn a background worker from the unicorn process changed. Run
heroku config:remove SERVER_EMBED_RESQUE_WORKER
heroku config:set SERVER_EMBED_SIDEKIQ_WORKER=true
We're automatically adjusting the ActiveRecord connection pool size for you.
Larger Heroku setups should have enough expertise to figure out what to do
by them self.
The Capistrano deployment scripts were removed from the main source code
repository, since they were no longer working.
They will be moved into their own repository with a new maintainer,
you'll be able to find them under the Diaspora* Github organization once
everything is set up.
Published by jhass about 11 years ago
Published by jhass about 11 years ago
to
field blank when sending a private message causes a server error #4227
Published by jhass about 11 years ago
Attention: This release includes a potentially long running migration! However it should be safe to run this while keeping your application servers on.