Skip to content
This repository has been archived by the owner on Jan 13, 2025. It is now read-only.

A downed node appears to have too large of an impact on the blocks produced by the next leader in line #7588

Closed
mvines opened this issue Dec 20, 2019 · 1 comment · Fixed by #7764
Assignees

Comments

@mvines
Copy link
Contributor

mvines commented Dec 20, 2019

On an idling cluster running v0.21.5 with one validator offline, solana show-block-production -v shows that typically the leader following the offline validator misses two slots while it waits for the offline leader to timeout:

  2121738         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG
  2121739         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG
  2121740         2te46rxywMdCNdkvjumiBBPQoVczJFxhxEaxFavQNqe3 MISSED
  2121741         2te46rxywMdCNdkvjumiBBPQoVczJFxhxEaxFavQNqe3 MISSED
  2121742         2te46rxywMdCNdkvjumiBBPQoVczJFxhxEaxFavQNqe3 MISSED
  2121743         2te46rxywMdCNdkvjumiBBPQoVczJFxhxEaxFavQNqe3 MISSED
  2121744         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG MISSED
  2121745         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG MISSED
  2121746         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG
  2121747         5BD5PWT7Nn3ZFxZFXKgbrzFZxPi7FxxXHyEdbmVuPEGG

Consider reducing grace ticks?

@mvines mvines added this to the Supertubes v0.22.1 milestone Dec 20, 2019
@pgarg66
Copy link
Contributor

pgarg66 commented Dec 20, 2019

We can also test if the previous leader (2te46rxywMdCNdkvjumiBBPQoVczJFxhxEaxFavQNqe3 in example above) has not transmitted any data in its leader slots. If so, the grace ticks can be ignored by the next leader. This will remove any delays due to missing nodes.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

Successfully merging a pull request may close this issue.

2 participants