>_DevOps & deployment
Follow CI, canary releases, rollback, and infrastructure automation into production. — 28 entries
Canary deployment카나리 배포
A new version receives a small share of traffic first, then more as its metrics are checked.
Continuous integration지속적 통합
Merge changes frequently and run automated checks to catch integration failures early.
Continuous delivery and deployment지속적 전달과 배포
Move a verified artifact through environment gates toward production.
Infrastructure as code코드형 인프라
Declare desired infrastructure in files, review the plan, and apply the difference.
Container image layers컨테이너 이미지 레이어
Build an image as reusable layers produced by successive instructions.
Kubernetes pod쿠버네티스 파드
A deployment unit whose containers share networking and storage.
Kubernetes service쿠버네티스 서비스
Expose a stable network address in front of replaceable pods.
Kubernetes reconciliation loop쿠버네티스 조정 루프
A controller repeatedly closes the gap between desired and actual state.
Horizontal autoscaling수평 자동 확장
Change the number of running replicas in response to load metrics.
Blue-green deployment블루·그린 배포
Prepare old and new environments side by side, then switch the entry route.
Feature flag release기능 플래그 출시
Separate deployment from exposure and enable a feature for selected cohorts.
GitOps reconciliationGitOps 조정
Detect and reconcile differences between a Git declaration and the running environment.
Rollback롤백
Return traffic or the running version to a previous release after a problem.
Dynamic resource allocation동적 자원 할당
Match a pod request to a suitable device through resource claims.
Mutating admission policy변경형 어드미션 정책
An API server policy fills or changes resource fields before persistence.
Pod user namespaces파드 사용자 네임스페이스
Map container user IDs to different, unprivileged host IDs.
Pressure stall metrics압력 정체 지표
Measure time spent waiting for resources even when utilization looks similar.
OCI artifact volumeOCI 아티팩트 볼륨
Pull image or artifact content from a registry and mount it as a pod volume.
Workload-aware scheduling워크로드 인식 스케줄링
Schedule related pods as a group when they need to start together.
2026 SBOM minimum elements2026 SBOM 최소 요소
Record component hashes, licenses, tool names, and generation context under the updated 2026 guidance.
Node log query노드 로그 조회
Read node system logs through the kubelet API without an SSH session.
External ServiceAccount token signer외부 서비스 계정 토큰 서명기
Move ServiceAccount token signing to a key system outside the API server.
HPA scale to zeroHPA 제로 스케일
Use an external metric to scale idle workloads to zero pods and back.
Allocated device health할당 장치 상태
Read the health of assigned devices such as GPUs from pod status.
Volume group snapshots볼륨 그룹 스냅샷
Capture a crash-consistent point in time across multiple persistent volumes.
Platform engineering플랫폼 엔지니어링
Give teams self-service templates for standard infrastructure and delivery paths.
Preview environment프리뷰 환경
Create a temporary running environment for each change request and remove it on close.
Deployment freeze배포 동결
Pause ordinary production releases during a risky period while allowing reviewed emergencies.