Closed bhalevy closed 5 months ago
The relevant code that checks the "validity" of a system.peers
row:
https://github.com/scylladb/python-driver/blob/6b82872ef0bbd1fc9cf78fb069a7d7821e3a6dde/cassandra/cluster.py#L4018-L4021
and the warning printed here (https://github.com/scylladb/python-driver/blob/6b82872ef0bbd1fc9cf78fb069a7d7821e3a6dde/cassandra/cluster.py#L3949-L3953) should print more details (entire row?) and which of the cases caused the invalidity of a the row.
It looks like the error heals itself eventually but it's unclear why the the row in system.peers was considered as invalid, and on which node it was read from.
I addressed "why the the row in system.peers was considered as invalid" part in #305, however not necessarily the "and on which node it was read from" part but that can be deduced from other logs.
It looks like the error heals itself eventually but it's unclear why the the row in system.peers was considered as invalid, and on which node it was read from.
I addressed "why the the row in system.peers was considered as invalid" part in #305, however not necessarily the "and on which node it was read from" part but that can be deduced from other logs.
thanks. hopefully it will be enough. heads up @kbr-scylla
I'm seeing these error in a dtest that adds 2 new nodes (in parallel) to a 3-nodes cluster.
It looks like the error heals itself eventually but it's unclear why the the row in system.peers was considered as invalid, and on which node it was read from. It would help if we printed more information to facilitate debugging in case something is wrong in the scylla server side.