`unit/cluster/faster-failover.tcl` has been in the tree failing since the
merge: three tests, two asserting on a log line no source emits and the
third dying on `invalid command name "memcmp"`. Same cause as MSETEX and
CLUSTERSCAN — the new test file was taken, the edits to existing files
were not.
Recovered from upstream:
ba076abd4 Do the failover immediately if the replica is the best
ranked replica (#2227) — cluster_legacy.{c,h}, and the
`memcmp` helper in tests/support/util.tcl that #2227 added
for exactly this test
38642302b Trigger failover check ASAP when cluster-replica-no-failover
is disabled (#3827)
`memcmp` is worth noting: it is the one helper the rebrand could not have
renamed, because its name carries no brand. Nothing pointed at it, so it
was simply dropped — a reminder that the rename audit and the content
audit are different audits.
Also fixed `$$` where `&&` belongs at faster-failover.tcl:297, which made
`expr` fail to parse and aborted the suite. Introduced upstream in
b6e3e5339 (#3495); the two conditions immediately above it are identical
and correct.
Tests: unit/cluster/faster-failover 3/3, was 0/3 with an abort.
Co-authored-by: Hanzo Dev <[email protected]>