Michael Penick
Michael Penick
`cassandra.yaml` settings load by the system property `stargate.unsafe.cassandra_config_path` will be overridden by also providing `--emulate-dbaas-defaults`. https://github.com/stargate/stargate/blob/v1.0.54/persistence-dse-6.8/src/main/java/io/stargate/db/dse/DsePersistenceActivator.java#L146-L150 ```java if (Boolean.parseBoolean(System.getProperty("stargate.emulate_dbaas_defaults", "false"))) { c.emulate_dbaas_defaults = true; GuardrailsConfig guardrailsConfig = new GuardrailsConfig(); guardrailsConfig.secondary_index_per_table_failure_threshold...
We should remove (or severely deprecate) overloads of `Persistence.newConnection()` that don't require a `ClientInfo` with a valid authenticated user. Maybe this isn't possible while supporting the C* backends in persistence?...
Example: Query with page size 2000 (REST, GraphQL), receive 1618 rows. Total number available rows at the current position is 315k+ so this should have returned 2000 rows. It's possible...
This builds on an existing PR: https://github.com/graphql-go/graphql/pull/401. It fixes issues found in that PR (it works for enums, objects, and lists values) and adds tests. Externally, this uses `nil` instead...
It looks like gocql added support for `peers_v2` and doesn't fallback if it isn't present in some cases. https://github.com/gocql/gocql/commit/8f058c60165fc3be0e4e483966fbbe76186e39b8
Run performance tests to quantify the impact of the proxy versus connecting directly to a cluster. Also, quantify the saturation point in terms of throughput and number of connections. Essentially,...
If an unavailable or read/write timeout are already handled by the proxy it shouldn't just forward those to driver as it might cause the retry policy to retry again. Maybe...