1.7. Upgrade duration and impact


The durations in the following table were recorded in a test environment that consisted of an overcloud with 200 nodes, and 9 Ceph Storage hosts with 17 object storage daemons (OSDs) each. The durations in the table might not apply to all production environments. For example, if your hardware has low specifications or an extended boot period, allow for more time with these durations. Durations also depend on network I/O to container and package content, and disk I/O on the host.

To accurately estimate the upgrade duration for each task, perform these procedures in a test environment with hardware that is similar to your production environment.

Expand
표 1.3. Duration and impact of In-place upgrade
 DurationNotes

Undercloud upgrade

  • 30 minutes
  • No disruption to the overcloud.

Overcloud adoption and preparation

  • 10 minutes for bare metal adoption
  • 30 minutes for upgrade prepare
  • Duration scales based on the size of the overcloud.
  • No disruption to the overcloud.

Red Hat Ceph Storage upgrade

  • Ceph upgrade ansible run: 90 minutes total, 10 minutes per node
  • Ceph ansible run for cephadm adoption: 30 minutes total, 3 minutes per node
  • Post ceph upgrade and adoption overcloud upgrade prepare: 20 minutes
  • Duration scales based on the number of Red Hat Ceph Storage hosts and OSDs.
  • Storage performance is degraded.

Overcloud OpenStack upgrade

  • 120 minutes
  • Duration scales based on the size of the overcloud.
  • During this process, agents are restarted and API transactions might be lost. Disable client access to the OpenStack API during this stage.
  • Workloads that are running in the overcloud remain active.
  • The upgrade does not affect the following:

    • Red Hat Ceph Storage OSDs (back-end storage for instances)
    • Network connectivity (East-West, North-South)
    • Undercloud

Undercloud system upgrade

  • 40 minutes
  • Includes multiple reboots. Reboot times are hardware dependent.
  • Includes SELinux relabeling. Hosts with large numbers of files take significantly longer.
  • No disruption to the overcloud.

Overcloud non-Compute host system upgrade

  • 30 minutes for upgrade prepare
  • 40 minutes per host system upgrade
  • Includes multiple reboots. Reboot times are hardware dependent.
  • Workloads that are running in the overcloud remain active.
  • Includes SELinux relabeling. Hosts with large numbers of files take significantly longer.
  • Performance is degraded.

Overcloud Compute host upgrade

  • 40 minutes per batch of hosts
  • 30 minutes for upgrade prepare
  • You run upgrade prepare on select batches of Compute nodes. The duration depends on the number of Compute nodes in each batch. There is no outage.
  • Includes multiple reboots. Reboot times are hardware dependent.
  • Includes SELinux relabeling. Hosts with large numbers of files take significantly longer.
  • To prevent workload outages during the reboot, you can migrate the workloads to another host beforehand.
Red Hat logoGithubredditYoutubeTwitter

자세한 정보

평가판, 구매 및 판매

커뮤니티

Red Hat 소개

Red Hat은 기업이 핵심 데이터 센터에서 네트워크 에지에 이르기까지 플랫폼과 환경 전반에서 더 쉽게 작업할 수 있도록 강화된 솔루션을 제공합니다.

보다 포괄적 수용을 위한 오픈 소스 용어 교체

Red Hat은 코드, 문서, 웹 속성에서 문제가 있는 언어를 교체하기 위해 최선을 다하고 있습니다. 자세한 내용은 다음을 참조하세요.Red Hat 블로그.

Red Hat 문서 정보

Legal Notice

Theme

© 2026 Red Hat
맨 위로 이동