glesica/mtdata

Add basic retry with modest back-off

glesica opened this issue · 0 comments

When a fetch operation fails, we should retry a configurable number of times with some kind of very modest back-off (to keep the length of a run short). For example, if it fails once, we should wait at least 5 seconds and then try again. If it fails again, we should wait, say, 10 seconds and then try again.

We could also consider retrying store operations, but fetch is the thing most likely to fail.