headscale/hscontrol/state
Kristoffer Dalby fb8eecae25 state: defer HA failover when probe target reconnected mid-cycle
The HA prober dispatches a PingRequest, waits ProbeTimeout (5s), and
marks the node unhealthy if no callback arrives. A node that bounced
its poll session between probe cycles satisfies two conditions that
conspire to fail TestHASubnetRouterFailover: a probe queued against
the previous session is silently dropped when the worker writes to
the closed connection (timeout always fires), and a probe sent
immediately after reconnect lands while wgengine is still rebuilding
magicsock state from the new netmap. Either path installs a spurious
unhealthy bit, which sends the preserved-primary anti-flap the wrong
way.

Record the session observed at dispatch time and drop the timeout
path if the node reconnected since. Require the session to survive
a full probe cycle before a timeout can drive a failover.
2026-05-18 17:18:08 +02:00
..
auth_cache_test.go state: replace zcache with bounded LRU for auth cache 2026-04-10 14:09:57 +01:00
debug.go state: switch consumers to NodeStore primary routes 2026-04-29 18:08:39 +01:00
debug_test.go hscontrol/state: make NodeStore batch configuration tunable (#2886) 2025-11-28 16:38:29 +01:00
endpoint_test.go hscontrol/state: make NodeStore batch configuration tunable (#2886) 2025-11-28 16:38:29 +01:00
ephemeral_test.go all: upgrade to Go 1.26rc2 and modernize codebase 2026-02-08 12:35:23 +01:00
ha_health.go state: defer HA failover when probe target reconnected mid-cycle 2026-05-18 17:18:08 +02:00
maprequest.go all: fix golangci-lint issues (#3064) 2026-02-06 21:45:32 +01:00
maprequest_test.go all: fix golangci-lint issues (#3064) 2026-02-06 21:45:32 +01:00
node_store.go state: preserve previous primary when all HA advertisers unhealthy 2026-04-29 18:08:39 +01:00
node_store_hostname_test.go state: auto-bump GivenName on collision and add SetGivenName 2026-04-18 15:12:21 +01:00
node_store_test.go types: persist Node JSON slices via named IsZero types 2026-05-15 11:21:58 +02:00
persist_test.go state: add regression test for Node slice persistence 2026-05-15 11:21:58 +02:00
ping.go state: drain pending pings on Close 2026-04-17 16:31:49 +01:00
ping_test.go state: drain pending pings on Close 2026-04-17 16:31:49 +01:00
primaries_property_test.go state: preserve previous primary when all HA advertisers unhealthy 2026-04-29 18:08:39 +01:00
primaries_test.go state: preserve previous primary when all HA advertisers unhealthy 2026-04-29 18:08:39 +01:00
ssh_check_test.go state, policy, noise: implement SSH check period auto-approval 2026-02-25 21:28:05 +01:00
state.go mapper, state: deliver nodeAttrs through MapResponse and harden nextdns DoH rewrite 2026-05-13 14:22:30 +02:00
tags.go hscontrol: enforce that tagged nodes never have user_id 2026-02-20 21:51:00 +01:00
test_helpers.go hscontrol/state: make NodeStore batch configuration tunable (#2886) 2025-11-28 16:38:29 +01:00