You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This came up while writing up the more general "stellar-archivist" issue stellar/go#4549
Basically: when replaying from history nodes are not aware of SCP messages.
For validators:
this creates situations where what the node is going to publish to history does not contain SCP data at all or has gaps.
For watcher nodes:
by not using SCP messages, the meta will not contain SCP data at all. This may create problems in downstream systems (like bridges) that need to consume SCP messages for all ledgers (as "walking back" the ledger chain from a trusted ledger is a lot more expensive).
A few things to consider:
it would be nice if we can recover SCP messages from other archives in some way during catchup
we can't assume that a given archive is complete, so we have to potentially download from multiple archives
quorum set configuration changes over time, so it may be hard to tell if we have all SCP messages, or if what we have is a superset of what we're supposed to publish (special consideration will have to be taken when dealing with archives that can't be fully trusted, maybe having a way to tell in config which archives can be used for that purpose would make sense?).
The text was updated successfully, but these errors were encountered:
This came up while writing up the more general "stellar-archivist" issue stellar/go#4549
Basically: when replaying from history nodes are not aware of SCP messages.
For validators:
this creates situations where what the node is going to publish to history does not contain SCP data at all or has gaps.
For watcher nodes:
by not using SCP messages, the meta will not contain SCP data at all. This may create problems in downstream systems (like bridges) that need to consume SCP messages for all ledgers (as "walking back" the ledger chain from a trusted ledger is a lot more expensive).
A few things to consider:
it would be nice if we can recover SCP messages from other archives in some way during catchup
we can't assume that a given archive is complete, so we have to potentially download from multiple archives
quorum set configuration changes over time, so it may be hard to tell if we have all SCP messages, or if what we have is a superset of what we're supposed to publish (special consideration will have to be taken when dealing with archives that can't be fully trusted, maybe having a way to tell in config which archives can be used for that purpose would make sense?).
The text was updated successfully, but these errors were encountered: