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
Since 2ec1f71 nodes no longer publish HIBERNATE state. This commit ended up in Scylla 4.6.
However, the old code for handling HIBERNATE state is still there:
} else if (move_name == sstring(versioned_value::HIBERNATE)) {
co_await handle_state_replacing(endpoint);
We could remove this and replace it with an ERROR message if a node publishes HIBERNATE - that is indeed an error, since it would mean we're doing replace in a mixed cluster - and with a <=4.5 node replacing one of our nodes (which would probably fail earlier anyway since the node would not pass feature check).
The text was updated successfully, but these errors were encountered:
Documentation must also be updated: docs/operating-scylla/procedures/cluster-management/replace-dead-node.rst mentions that the replacing node goes through hibernate state, which is wrong.
Since 2ec1f71 nodes no longer publish HIBERNATE state. This commit ended up in Scylla 4.6.
However, the old code for handling HIBERNATE state is still there:
We could remove this and replace it with an ERROR message if a node publishes HIBERNATE - that is indeed an error, since it would mean we're doing replace in a mixed cluster - and with a <=4.5 node replacing one of our nodes (which would probably fail earlier anyway since the node would not pass feature check).
The text was updated successfully, but these errors were encountered: