Prepare for ScyllaDB move to host ids instead of ips #645
Conversation
Current host_id(0 implementation works only if a node is UP since it contacts the node itself. Change it so that the id is looked up in the logs and only if this fails we try to contact the node. This way we can check host id even for a node that is down.
We are changing ScyllaDB to use ids internally, so some logs are changing to use ids instead of ips. Prepare for that.
|
@gleb-cloudius FYI after this would be review/merged, the sub-module in dtest would need to be updated. also I've trigger all of the gating tests with it, to validate it's not breaking anything. |
|
Restarted the first one: https://jenkins.scylladb.com/job/scylla-staging/job/dtest-pytest-gating/897/ |
It's was a test known to be unstable, and the rerun passed |
Now we can do the dtest part |
|
This needs to be backported to 2025.1, as it seems to be failing on it as well, I suspect. |
the dtest PR can be backported, once someone would open it up |
https://github.com/scylladb/scylla-dtest/pull/5699, marked it for backport |
3 participants
Moving ScyllaDB to host ids changes some output to use ids instead of ips. Prepare for that by checking for both.
Testing