. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 17.2. Equality tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 17.2.3. Summary of (in)equality tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . result can be understood by writing out the entire expression as a disjunction (OR-chain) of equality tests: A=Expr1 or A=Expr2 or ... or A=ExprN which, if A is NULL, boils down to NULL or NULL or ... or

startup. With both auto-retry mechanisms disabled, TiDB 2.1.8 through 3.0.0-beta.1-40 passed our tests for snapshot iso- lation and single-key linearizability. 3.0.0-rc.2, which disables auto-retry by some invariants. We tested TiDB 2.1.7, 2.1.8, 3.0.0-beta.1, 3.0.0-beta.1- 40, and 3.0.0-rc.2. Our tests ran on a five-node De- bian Stretch cluster, with PD, TiKV, and TiDB on each node. We used a replication and CLOCK_MONOTONIC, running up to 5x faster than oth- ers. PingCAP had written their own Jepsen tests, which we reviewed and expanded upon during our collabo- ration. The resulting test suite encompasses

............... 873 33. Regression Tests ................................................................................................ 875 33.1. Running the Tests .................................. Comparison Files .......................................................................... 882 33.4. TAP Tests .............................................................................................. 883 operator precedence rules. In particular, <= >= and <> used to be treated as generic operators; IS tests used to have higher priority; and NOT BETWEEN and related constructs acted inconsistently, being taken

..............................................................420 16.1.5. Running the Regression Tests..................................................................421 16.1.6. Building the Documentation .......625 30. Regression Tests................................................................................................................627 30.1. Running the Tests ........................... ..........................627 30.1.1. Running the Tests Against a Temporary Installation...............................627 30.1.2. Running the Tests Against an Existing Installation .................

..............................................................446 16.1.5. Running the Regression Tests.........................................................................447 16.1.6. Building the Documentation ....662 30. Regression Tests.......................................................................................................................663 30.1. Running the Tests ....................... ......................663 30.1.1. Running the Tests Against a Temporary Installation......................................663 30.1.2. Running the Tests Against an Existing Installation ..............

................. 832 33. Regression Tests ........................................................................................... 833 33.1. Running the Tests ..................................... Comparison Files .................................................................... 839 33.4. TAP Tests ......................................................................................... 840 33.5 operator precedence rules. In particu- lar, <= >= and <> used to be treated as generic operators; IS tests used to have higher pri- ority; and NOT BETWEEN and related constructs acted inconsistently, being

............... 841 32. Regression Tests ................................................................................................ 843 32.1. Running the Tests .................................. Comparison Files .......................................................................... 849 32.4. TAP Tests .............................................................................................. 850 operator precedence rules. In particular, <= >= and <> used to be treated as generic operators; IS tests used to have higher priority; and NOT BETWEEN and related constructs acted inconsistently, being taken

................. 804 32. Regression Tests ........................................................................................... 805 32.1. Running the Tests ..................................... Comparison Files .................................................................... 811 32.4. TAP Tests ......................................................................................... 812 32.5 operator precedence rules. In particular, <= >= and <> used to be treated as generic operators; IS tests used to have higher priority; and NOT BETWEEN and related constructs acted inconsistently, being taken

................. 801 32. Regression Tests ........................................................................................... 802 32.1. Running the Tests ..................................... Comparison Files .................................................................... 808 32.4. TAP Tests ......................................................................................... 809 32.5 operator precedence rules. In particular, <= >= and <> used to be treated as generic operators; IS tests used to have higher priority; and NOT BETWEEN and related constructs acted inconsistently, being taken

............... 847 32. Regression Tests ................................................................................................ 849 32.1. Running the Tests .................................. Comparison Files .......................................................................... 855 32.4. TAP Tests .............................................................................................. 856 operator precedence rules. In particular, <= >= and <> used to be treated as generic operators; IS tests used to have higher priority; and NOT BETWEEN and related constructs acted inconsistently, being taken