. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 310 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . the change introduces. The thumb rule is: if it can happen in production, it must be covered. • Integration tests should ensure that the feature works end-to-end. They must cover all the important use cases transaction support to REPLICA (or other tablet types). Use Case(s) • Consistent reads • Sqoop integration 48 Proposed Solution • When vtgate chooses a tablet to execute a query on, it should return

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 333 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . the change introduces. The thumb rule is: if it can happen in production, it must be covered. • Integration tests should ensure that the feature works end-to-end. They must cover all the important use cases vtgate on PRIMARY and read REPLICA tablets as of v7.0. Use Case(s) • Consistent reads • Sqoop integration Implimented Solution • When vtgate chooses a tablet to execute a query on, it will return the

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . the change introduces. The thumb rule is: if it can happen in production, it must be covered. • Integration tests should ensure that the feature works end-to-end. They must cover all the important use cases transaction support to REPLICA (or other tablet types). Use Case(s) • Consistent reads • Sqoop integration Proposed Solution • When vtgate chooses a tablet to execute a query on, it should return the

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . the change introduces. The thumb rule is: if it can happen in production, it must be covered. • Integration tests should ensure that the feature works end-to-end. They must cover all the important use cases keyspaces for recovery of the last backup before a requested specific timestamp to restore to. 2. Integration with a binlog server to allow vttablet to apply binary logs from the recovered backup up to the

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . change introduces. The thumb rule is: if it can happen in production, it must be covered. • Integration tests should ensure that the feature works end-to-end. They must cover all the important use cases transaction support to REPLICA (or other tablet types). Use Case(s) • Consistent reads • Sqoop integration Proposed Solution • When vtgate chooses a tablet to execute a query on, it should return the

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . proper unit test coverage for this change? All changes should increase coverage. We need at least integration test coverage when unit test coverage is not possible. 23 • Is this change going to log too much typically have 3 or 5 servers, in different server racks / sub-networks for higher resilience. For our integration tests, we use a single ZK server that serves both global and local cells. We provide the zk utility

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . proper unit test coverage for this change? All changes should increase coverage. We need at least integration test coverage when unit test coverage is not possible. • Is this change going to log too much typically have 3 or 5 servers, in different server racks / sub-networks for higher resilience. For our integration tests, we use a single ZK server that serves both global and local cells. We provide the zk utility

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195 Integration with Orchestrator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . proper unit test coverage for this change? All changes should increase coverage. We need at least integration test coverage when unit test coverage is not possible. • Is this change going to log too much typically have 3 or 5 servers, in different server racks / sub-networks for higher resilience. For our integration tests, we use a single ZK server that serves both global and local cells. We provide the zk utility

2 CVEs assigned Formalisation of VTAdmins threat model 3 fuzzers added to Vitessʼs OSS-Fuzz integration 2 Vitess Security Audit, 2023 Notable findings The most notable findings from the audit are were found during the manual auditing. Ada Logics added the three fuzzers to Vitess's OSS-Fuzz integration, allowing them to run continuously and test for more special cases as well as code changes. 15

minikube-instructions. This was done to gain better understanding of the general deployment structure and the integration with the core components. • The initially enumerated application endpoints were tested for potential