While we don't really need to worry about retries for our keyring implementation or Clevis with TPM2, we should consider taking the CoreOS team's advice to make it slightly more resistant to network failures.
@mulkieran I think the exponential back off may make sense but we have a problem here because if we wait too long to do retries, we hang the boot. This is mostly a question of balancing user experience vs. making the set up resilient to network failures. Do you have any thoughts here?
While we don't really need to worry about retries for our keyring implementation or Clevis with TPM2, we should consider taking the CoreOS team's advice to make it slightly more resistant to network failures.
@mulkieran I think the exponential back off may make sense but we have a problem here because if we wait too long to do retries, we hang the boot. This is mostly a question of balancing user experience vs. making the set up resilient to network failures. Do you have any thoughts here?