Ceph Squid to Tentacle upgrade
Upgrading a cephadm-managed Squid cluster to Tentacle before Squid's estimated end of life: health checks, staggered daemon upgrade order, require-osd-release, new defaults and a rollback-free test plan.
Ceph upgrades
Upgrade pages for moving between Ceph major releases with cephadm, based on the Ceph release notes and releases index.
Ceph Reef to Tentacle upgrade
Taking an out-of-support Reef cluster straight to Tentacle, which the Tentacle notes support: prerequisites, conversion of older non-cephadm clusters, upgrade sequencing and post-upgrade checks.
Ceph Quincy to Squid upgrade
Moving a Quincy cluster forward when Tentacle does not support it as a source: upgrading to Squid first, then on to Tentacle before Squid's estimated end of life, with checks between the two hops.
Migrating FileStore OSDs to BlueStore before upgrading
FileStore OSDs cannot run on Reef or later: choosing mark-out replacement, whole-host replacement or per-OSD copy, keeping destruction inside one failure domain, and sequencing conversion with the release upgrade.
Ceph CRUSH tunables and client compatibility after upgrades
Handling the non-optimal tunables warning after an upgrade: how much data moves, which older kernel and librados clients are cut off, and planning the change around client upgrades.
Upgrading Ceph in an air-gapped network
Running cephadm upgrades without internet access: mirroring the new Ceph and monitoring images into a local registry, pointing the upgrade at the local image and checking every host can pull it.
Applying Ceph point releases and security fixes
Keeping a supported release patched: which releases receive backports and security fixes, reading point release notes for CVE fixes, and rolling a point release through a cluster with cephadm.
