Originally from the User Slack
@Keith_Mciff: We have a cluster that completely lost all data drives. We have attached new storage and are trying to start the nodes but we are getting the below error when trying to start the first seed node. What file can we delete or command can we run to make this a “clean” first install again?
[shard 0] init - Shutting down sighup was successful
Jan 05 18:06:44 ip-XXXXXX.ec2.internal scylla[9518]: [shard 0] init - Startup failed: std::runtime_error (Failed to learn about other nodes' tokens during bootstrap. Make sure that:
- the node can contact other nodes in the cluster,
- the `ring_delay` parameter is large enough (the 30s default should be enough for small-to-middle-sized clusters),
- a node with this IP didn't recently leave the cluster. If it did, wait for some time first (the IP is quarantined),
and retry the bootstrap.)
@Felipe_Cardeneti_Mendes: 1. Ensure that the “first seed node” is the node with the “lowest IP address in your seed config. To make it simpler, you can just leave the first node to be the only seed entry and update it later
2. Follow https://opensource.docs.scylladb.com/stable/operating-scylla/procedures/cluster-management/clear-data.html
@Keith_Mciff: Ill give it a shot
That worked! Thank you!