planb-cassandra
planb-cassandra copied to clipboard
zalando-stups
Plan B Cassandra for STUPS/AWS with static IPs
Due to change in data directory layout in older versions of planb-cassandra, while updating the docker image a node may lose its system keyspace. What happens in this case is...
And leads to the following problem (i.e. an endless loop at the last step): ``` 2017-05-18 13:05:25,477 INFO: Waiting for new instance to attach vol-1234abc channel 2: open failed: connect...
Concurrent update operations are not supported and probably not so easy to detect and guard against. We should at least document that to warn users against acting on the same...
Currently, any commit to the master branch triggers build if new docker images, most of them will be identical if we only change the create/update cluster script.
Currently `create` command produces colorful output by using click, while update is using `logger.INFO` mostly. Using click has some benefits, e.g. showing waiting progress in a more compact way.
Currently the failed state needs to be cleaned manually. We should prompt the user and do this automatically.
We want to be able to start with a specific node without need to explicitly skip others. Resume should also be smarter.
When performing a rolling update of a cluster, we terminate instances and create new ones, but in the process we may loose SNS notification topic if it was set previously...
It would be nice to have some documentation which steps need to be done in order to delete planb cassandra again.
In order to be able to authenticate with new created users after upgrade you actually have to drop the following tables: - system_auth.users - system_auth.credentials - system_auth.permissions Otherwise your cluster...
Metadata
Owner
Metadata
Plan B Cassandra for STUPS/AWS with static IPs