Guaranteed delivery of crash reports


my understanding so far is that the ideology is to attempt to deliver a crash report to Sentry, but if that is not possible (e.g. the system where the raven client runs cannot temporarily connect to the Sentry server) - just move on.

When that is the case, raven will write a message to the local logger instead (I got this feedback from someone on the IRC channel, when asking about raven for Python) and subsequently not reattempt a delivery.

  1. Is there a way to adjust the behaviour of the client such that it retries to send the crash report until it is delivered?
  2. Are there any ideological constraints that make the creation of reliable transports unreasonable or unnecessary?

I look forward to your comments, as I am not yet sure I fully grasped the best practices of using Sentry.