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
If this kind of failure persists for long enough and we have to wait for the timeout of 30secs to skip to the next immediate oracle for every query things stop working.
The solution is for the user to do the change themselves (change the order of oracles, or even delete the misbehaving oracle) but it would be nice if we could detect this and fix it for them automatically.
Specification
The current (and maybe historicall too?) price oracle manager class should keep a rolling record of queries per oracle and if they fail or succeed. If for an oracle all the X latest queries are failing then we should temporarily ban it
The manager should unban oracles after Y minutes have passed.
The text was updated successfully, but these errors were encountered:
Abstract
We have had the problem that a price oracle goes offline for a very long time such as coingecko for over 12 hours here: https://twitter.com/rotkiapp/status/1527554874191368193
If this kind of failure persists for long enough and we have to wait for the timeout of 30secs to skip to the next immediate oracle for every query things stop working.
The solution is for the user to do the change themselves (change the order of oracles, or even delete the misbehaving oracle) but it would be nice if we could detect this and fix it for them automatically.
Specification
The current (and maybe historicall too?) price oracle manager class should keep a rolling record of queries per oracle and if they fail or succeed. If for an oracle all the
X
latest queries are failing then we should temporarily ban itThe manager should unban oracles after
Y
minutes have passed.The text was updated successfully, but these errors were encountered: