Scylla test with all addresses ipv6 public fails on nodetool status #7447
Comments
|
also see errors like: |
|
Seems like we are comparing again without leading zeros, I thought I'd fixed it, but it seems we broke it again |
|
@juliayakovlev please take a look at this one, let's try to find why this isn't working as expected |
@fruch scylla-cluster-tests/sdcm/cluster_aws.py Line 463 in 3291578 |
o.k. so lets do (let try doing it on any place we retrieve ipv6 addresses, also for gce/azure ?) |
The problem was found during IPv6 test. IPv6 was converted to full format when parse nodetool status result. This change was represented by scylladb#7047 But IPv6 address was not converted when collected info about nodes network interfaces (AWS). During checking nodes status, we compare between 'nodetool status' output and node address that kept in the network_interfaces object. It fails because the IPv6 address miss leading zeros. As result the test fails with errors 'Failed to find a node in cluster by IP' Fixes: scylladb#7447
The problem was found during IPv6 test. IPv6 was converted to full format when parse nodetool status result. This change was represented by scylladb#7047 But IPv6 address was not converted when collected info about nodes network interfaces (AWS). During checking nodes status, we compare between 'nodetool status' output and node address that kept in the network_interfaces object. It fails because the IPv6 address miss leading zeros. As result the test fails with errors 'Failed to find a node in cluster by IP' Fixes: scylladb#7447
The problem was found during IPv6 test. IPv6 was converted to full format when parse nodetool status result. This change was represented by scylladb#7047 But IPv6 address was not converted when collected info about nodes network interfaces (AWS). During checking nodes status, we compare between 'nodetool status' output and node address that kept in the network_interfaces object. It fails because the IPv6 address miss leading zeros. As result the test fails with errors 'Failed to find a node in cluster by IP' Also fix the same problem for peer and gossip output. Fixes: scylladb#7447
The problem was found during IPv6 test. IPv6 was converted to full format when parse nodetool status result. This change was represented by #7047 But IPv6 address was not converted when collected info about nodes network interfaces (AWS). During checking nodes status, we compare between 'nodetool status' output and node address that kept in the network_interfaces object. It fails because the IPv6 address miss leading zeros. As result the test fails with errors 'Failed to find a node in cluster by IP' Also fix the same problem for peer and gossip output. Fixes: #7447
Packages
Issue description
Tried provision test with
configurations/network_config/all_addresses_ipv6_public.yamlconfig and failed onWaiting for nodes to join the cluster.While nodetool status shows all nodes up:
Installation details
Cluster size: 6 nodes (i4i.2xlarge)
Scylla Nodes used in this run:
OS / Image:
ami-03e1418be972f5f58(aws: undefined_region)Test:
provision-testTest id:
e6323f19-83c6-4ebc-97cb-23471b8b64feTest name:
scylla-staging/lukasz/provision-testTest config file(s):
Logs and commands
$ hydra investigate show-monitor e6323f19-83c6-4ebc-97cb-23471b8b64fe$ hydra investigate show-logs e6323f19-83c6-4ebc-97cb-23471b8b64feLogs:
Jenkins job URL
Argus
The text was updated successfully, but these errors were encountered: