adr: Staless based failover mechanism ADR proposed #208
Closed
johnride
wants to merge 1 commits from
adr/017-staleness-failover into master
pull from: adr/017-staleness-failover
merge into: NationTech:master
NationTech:master
NationTech:feat/openwebui
NationTech:feat/iot-walking-skeleton
NationTech:feat/iot-aggregation-scale
NationTech:feat/add-new-node
NationTech:feat/iot-operator-helm-chart
NationTech:feat/removesideeffect
NationTech:feat/test-alert-receivers-sttest
NationTech:feat/brocade-client-add-vlans
NationTech:feat/agent-desired-state
NationTech:feat/opnsense-dns-implementation
NationTech:feat/named-config-instances
NationTech:worktree-bridge-cse_012j1jB37XfjXvDGHUjHrKSj
NationTech:chore/leftover-adr
NationTech:feat/config_e2e_zitadel_openbao
NationTech:example/vllm
NationTech:feat/config_sqlite
NationTech:chore/roadmap
NationTech:feature/kvm-module
NationTech:feat/rustfs
NationTech:feat/harmony_assets
NationTech:feat/brocade_assisted_setup
NationTech:feat/cluster_alerting_score
NationTech:e2e-tests-multicluster
NationTech:fix/refactor_alert_receivers
NationTech:feat/change-node-readiness-strategy
NationTech:feat/zitadel
NationTech:feat/improve-inventory-discovery
NationTech:fix/monitoring_abstractions_openshift
NationTech:feat/nats-jetstream
NationTech:adr-nats-creds
NationTech:feat/st_test
NationTech:feat/dockerAutoinstall
NationTech:chore/cleanup_hacluster
NationTech:doc/cert-management
NationTech:feat/certificate_management
NationTech:fix/nats_non_root
NationTech:feat/rebuild_inventory
NationTech:fix/opnsense_update
NationTech:feat/unshedulable_control_planes
NationTech:feat/worker_okd_install
NationTech:doc-and-braindump
NationTech:fix/pxe_install
NationTech:switch-client
NationTech:okd_enable_user_workload_monitoring
NationTech:configure-switch
NationTech:fix/clippy
NationTech:feat/gen-ca-cert
NationTech:feat/okd_default_ingress_class
NationTech:fix/add_routes_to_domain
NationTech:secrets-prompt-editor
NationTech:feat/multisiteApplication
NationTech:feat/ceph-install-score
NationTech:feat/ceph-osd-score
NationTech:feat/ceph_validate_health
NationTech:better-indicatif-progress-grouped
NationTech:feat/crd-alertmanager-configs
NationTech:better-cli
NationTech:opnsense_upgrade
NationTech:feat/monitoring-application-feature
NationTech:dev/postgres
NationTech:feat/cd/localdeploymentdemo
NationTech:feat/webhook_receiver
NationTech:feat/kube-prometheus
NationTech:feat/init_k8s_tenant
NationTech:feat/discord-webhook-receiver
NationTech:feat/kube-prometheus-monitor
NationTech:feat/tenantScore
NationTech:feat/teams-integration
NationTech:feat/slack-notifs
NationTech:monitoring
NationTech:runtime-profiles
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
No description provided.
Delete Branch "adr/017-staleness-failover"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Proposition d'une architecture pour gérer concrètement le failover entre deux sites.
On se fie au consensus de nats pour que le primary auto-valide qu'il fait encore partie de la majorité
Si une écriture nats échoue il se met immédiatement en read-only
Le replica vérifie son heartbeat timestamp, lorsqu'il va au-delà du timeout, il s'auto-promote selon la même contrainte qu'il doit être capable d'écrire, donc de faire partie de la majorité
Dès que le primary est de retour, le secondary s'auto-demote et le primary se promote
C'est l'architecture simplifiée ici, il y a des détails de plus à tenir en compte comme la gestion du temps de synchronisation des données entre les sites selon la stratégie de réplication utilisée. Mais c'est un bon point de départ.
Pull request closed