mirror of
https://github.com/taylanbakircioglu/haproxy-openmanager.git
synced 2026-10-23 20:45:41 +00:00
taylanbakircioglu
349434fc4d
A node could be a member of one VIP only: the form refused a second one, and the delivery endpoint served each agent a single VIP's keepalived.conf. - Delivery merges the node's applied VIPs into the one file it deploys: global_defs once, then per VIP a vrrp_script chk_haproxy_<id> (each keeps its own failover weight) and its vrrp_instance, ordered by VIP id. A node with one VIP gets the same bytes and hash as before, so existing nodes do not rewrite or reload. - One ack converges every VIP in the merged file; a deleted VIP's membership is marked disabled once the node runs a file without it. - A snapshot that cannot be merged leaves the node alone (status compose_error, members show the reason). A database error during delivery answers "unavailable" instead of "not_configured", which on a managed node meant teardown. - Adoption refuses a keepalived.conf with several vrrp_instance blocks (taking it over would remove the others), also for analyses stored by older versions and at Apply for an adoption made before the upgrade. Discoveries record the parser version; files analysed by an older one are re-reported once by v1.11.1+ agents. - Schema 13 is a delta from 12 (two additive columns); built-in roles are not re-seeded on a database already at 12. No agent change.