Ettore Di Giacinto 5ecaf8f792 fix(failover): count a chain switch once per cluster ...

Every frontend recorded localai_failover_switches_total for the switches
it adopted from the leader, so a cluster of N frontends counted each
switch N times. Only the leader, which decides the switch, records it.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Ettore Di Giacinto <mudler@localai.io>

2026-09-27 07:42:21 +00:00

..

distsync

fix(failover): re-sync pins on every frontend after a missed delta

2026-09-27 07:42:21 +00:00

classify_test.go

fix(failover): do not trip targets on a gRPC message size limit

2026-09-27 07:42:21 +00:00

classify.go

fix(failover): do not trip targets on a gRPC message size limit

2026-09-27 07:42:21 +00:00

failover_suite_test.go

feat(failover): add types and retryable error classification

2026-09-27 07:42:20 +00:00

fakes_test.go

fix(failover): answer HasChains from a flag set at sync

2026-09-27 07:42:20 +00:00

manager_test.go

fix(failover): trip a rate-limited or exhausted target instead of skipping it

2026-09-27 07:42:21 +00:00

manager.go

fix(failover): count a chain switch once per cluster

2026-09-27 07:42:21 +00:00

metrics_switch_test.go

fix(failover): count a chain switch once per cluster

2026-09-27 07:42:21 +00:00

metrics_test.go

feat(failover): run the chain manager and keep warm targets loaded

2026-09-27 07:42:20 +00:00

metrics.go

feat(failover): run the chain manager and keep warm targets loaded

2026-09-27 07:42:20 +00:00

prober_test.go

fix(failover): pass credential lookup from CLI

2026-09-27 07:42:20 +00:00

prober.go

fix(failover): pass credential lookup from CLI

2026-09-27 07:42:20 +00:00

schedule_test.go

fix(failover): never load a warm target inside a probe

2026-09-27 07:42:20 +00:00

schedule.go

fix(failover): re-sync pins on every frontend after a missed delta

2026-09-27 07:42:21 +00:00

statesync_test.go

fix(failover): roll back a pin that could not be shared

2026-09-27 07:42:21 +00:00

statesync.go

fix(failover): re-sync pins on every frontend after a missed delta

2026-09-27 07:42:21 +00:00

trace.go

feat(failover): run the chain manager and keep warm targets loaded

2026-09-27 07:42:20 +00:00

types_test.go

feat(failover): add types and retryable error classification

2026-09-27 07:42:20 +00:00

types.go

fix(failover): warn about warm on a remote target, align the spec

2026-09-27 07:42:20 +00:00