Diaspora* consistency

[Wouter Miltenburg]

Hi all,

Not sure if this is the right group to post this question so please excuse me if this is the case. However, I was wondering about consistency with Diaspora*. I read that sidekiq builds up a queue and will hold the messages that needs to be send. But what happens if a pod is temporarily down, will we still queue it and drop a message after a few attempts? What will happen if a node comes back online after a few days?

I also read that Redis is used as the key-value backend storage. However, it makes snapshots every two seconds. What will happen if Redis crashes and we lose some of the messages? Will friends still catch up these missed messages by fetching it from the upstream (e.g. not sending the message to your friends, but the friends picking up the messages by using an atom feed for example). Is there some kind of polling feature that will try to make the data consistent again?

With kind regards,

Wouter.