Changes since last release: https://github.com/github/orchestrator/compare/v3.0.12...v3.0.13

TL;DR Much Oracle GTID support, and many other things.

• GTID: detecting and analyzing errant transactions. On Oracle GTID clusters, orchestrator will:
  • Identify if a replica has errant transactions and will indicate the GTID set for those errant transactions, see https://github.com/github/orchestrator/pull/607
  • Web UI will visually indicate a server running with errant transactions, see https://github.com/github/orchestrator/pull/607
  • Allow "fixing" of errant transactions via reset master operation. This will auto calculate the errant range, and will correctly set gtid_executed and gtid_purged. orchestrator will protect against running this "fix" on an intermediate master. See https://github.com/github/orchestrator/pull/617
• GTID failure prediction:
  • Predict replication error and prevent relocation (https://github.com/github/orchestrator/pull/661). Will not allow relocating server A below server B if A is known to fail replication GTID-wise. MySQL can tell after effect, and orchestrator can tell before effect. This protects against loss of relay logs (especially important for SQL_DELAYed servers) and waste of futile operation. Thanks @MarkLeith, @evanelias, @sjmudd !
• GTID operations:
  • Better concurrency control, considers postponed-functions, utilizes failover optimizations (https://github.com/github/orchestrator/pull/641)
• Support for ReplicationCredentialsQuery config (https://github.com/github/orchestrator/pull/603), which allows orchestrator to grab replication credentials per-cluster; these can be used for setting up replication on graceful takeover or on master-master setup. Thanks @cezmunsta, @igroene !
• Fix to Zk KV update (https://github.com/github/orchestrator/pull/620)
• Fix to analysis GROUP BY statement (https://github.com/github/orchestrator/pull/632)
• Much internal refactoring (https://github.com/github/orchestrator/pull/623)
• Fixed integration tests on 5.7 (#649, #610), thanks @bbuchalter !
• Added web UI button for disable/enable global recoveries (#646, #656), thanks @mateusduboli !
• Added which-broken-replicas to list replicas with errors (#660), thanks @cezmunsta !
• Added which-cluster-osc-running-replicas command (#663), thanks @cezmunsta !
• take-siblings now uses "smart" logic (https://github.com/github/orchestrator/pull/670)
• raft internal health reports support HTTP auth. Thanks @almeida-pythian !
• Added MySQLConnectionLifetime, thanks @maciej-dobrzanski !
• Updates to docs, thanks @bbuchalter !
• Fix non-zero exit code for redeploy-internal-db (#624), thanks @bbuchalter !
• Include stdout/stderr in CommandRun's returned error (#635), thanks @bbuchalter !
• Fix: manual recovery overrides globally-disabled recoveries (#665)
• More visibility in logs in applying post-recovery steps (#666)

...and more!

$ gsha256sum *
2ca8f1d72ad1beae6048edc88efd3b0168593e42a2c6928e3cfa997d53062b79  orchestrator-3.0.13-1.x86_64.rpm
73ad98f312151c8aab5c708b940139fb8aa848c5891d096cf46ea83ce397bbe9  orchestrator-3.0.13-linux-amd64.tar.gz
9ceb60d669c2509f8d8d81f34591edcaafacf777606a995f742afcb1e40d3369  orchestrator-cli-3.0.13-1.x86_64.rpm
7b6f03b0b724e91d128b831ade51c1936efb847f0e39b9e749e370d59d631d7a  orchestrator-cli_3.0.13_amd64.deb
9192acf91ae0ba795918e20799fc6790af9275a3f25af73515e86cc5587dd826  orchestrator-client-3.0.13-1.x86_64.rpm
f2c13d0f679e8ef02216e2df697841f1ae4267a697c724901c0e5c5e09afa036  orchestrator-client_3.0.13_amd64.deb
26df8b44fa8bb6187f6b512a8fe4a63324b1faab006ec0c8959198a7b2e99427  orchestrator_3.0.13_amd64.deb