Skip to content

Document deploying Ceph with cephadm, and OSISM 11 - #1096

Open
ideaship wants to merge 4 commits into
mainfrom
osism-11-docs
Open

ideaship wants to merge 4 commits into
mainfrom
osism-11-docs

Conversation

@ideaship

@ideaship ideaship commented Oct 1, 2026 •

Copy link
Copy Markdown
Contributor

What

  • Deploy guide, Ceph: a "Deployment with cephadm" section, walking the osism apply cephadm-* plays in order — OSD device preparation with the osism-ansible LVM plays, bootstrap and the Ceph client, host registration, configuration, MON/MGR/crash and OSDs, pools and keys, CephFS/MDS and RGW, the dashboard, and the validators. It lists the prerequisites cfg-cookiecutter generates. The existing steps become "Deployment with ceph-ansible", for OSISM 10 and earlier and for clusters kept on ceph-ansible.
  • Release notes: docs/release-notes/osism-11.md, with what an operator has to know or do for OSISM 11, under "Next" until the release date is known. The index row is added with the release.

Notes for review

The step order is the one a full cephadm + Tentacle deployment was validated with. yarn build passes with no broken links or anchors.

The release notes describe changes from these, which have to land for the text to be true:

Part of

🤖 Generated with Claude Code

@github-actions

github-actions Bot commented Oct 1, 2026 •

Copy link
Copy Markdown

✅MegaLinter analysis: Success

Descriptor Linter Files Fixed Errors Max errors Warnings Elapsed time
✅ ACTION actionlint 5 0 0 0.05s
✅ JSON jsonlint 4 0 0 0.08s
✅ JSON prettier 4 0 0 0.35s
✅ JSON v8r 4 0 0 7.65s
✅ MARKDOWN markdownlint 172 0 0 3.35s
✅ MARKDOWN markdown-table-formatter 172 0 0 0.39s
✅ REPOSITORY betterleaks yes no no 1.22s
✅ REPOSITORY checkov yes no no 17.87s
✅ REPOSITORY git_diff yes no no 0.08s
✅ REPOSITORY secretlint yes no no 1.5s
✅ REPOSITORY trufflehog yes no no 4.1s
✅ SPELL codespell 182 0 0 0.64s
✅ SPELL lychee 182 0 0 14.34s
✅ YAML prettier 6 0 0 0.44s
✅ YAML v8r 6 0 0 6.82s
✅ YAML yamllint 6 0 0 0.5s

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

  • Documentation: Custom Flavors
  • Command: npx mega-linter-runner@10.1.0 --custom-flavor-setup --custom-flavor-linters ACTION_ACTIONLINT,JSON_JSONLINT,JSON_V8R,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,REPOSITORY_CHECKOV,REPOSITORY_GIT_DIFF,REPOSITORY_BETTERLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_TRUFFLEHOG,SPELL_LYCHEE,SPELL_CODESPELL,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

MegaLinter is provided by OX Security
Show us your support by starring ⭐ the repository

From OSISM 11 a new deployment deploys Ceph with cephadm, and the deploy
guide described ceph-ansible only. Add a section that walks the cephadm
plays in order: OSD device preparation with the osism-ansible LVM plays,
bootstrap and the Ceph client, host registration, configuration, the
MON/MGR/crash services and the OSDs, pools and keys, CephFS/MDS and RGW,
and the dashboard, followed by the validators.

It states the prerequisites that cfg-cookiecutter generates (the
dedicated cephadm SSH key, no ceph-ansible container, ceph_version in
the global configuration on the latest track). The existing steps become
"Deployment with ceph-ansible", for OSISM 10 and earlier and for
clusters kept on ceph-ansible; the RGW integration into OpenStack and
the key persistence are shared and referenced from the new section.

The order is the one a full cephadm Tentacle deployment was validated
with.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Roger Luethi <luethi@osism.tech>
Start the OSISM 11 release notes with what an operator has to know or
do: new installations only, OpenStack 2026.1, Ceph deployed with cephadm
on Tentacle and what that changes in configuration and commands, the
aes256k cephx keys and the kernel client limitation they bring, Ceph
validators that now fail on a failed validation, and the 2026.1 changes
operators touch. Entries sit under "Next" until the release date is
known; the series is added to the index with the release.

Assisted-by: Claude:claude-opus-5-5
Signed-off-by: Roger Luethi <luethi@osism.tech>
@ideaship ideaship self-assigned this Oct 2, 2026
OSISM 11 refuses a deployment whose host names are inconsistent, which
is a new way for a bootstrap to stop. An operator who hits it sees only
the failure message, and the names are decided before the deployment
runs, so the note has to be readable by someone who has not started
yet.

Describe the condition without the implementation: the name the kernel
reports against the name a lookup returns, case, resolvability, and
collisions across the inventory. Name the three lookups that break,
since "some things disagree" does not convey that each one is fatal on
its own.

State that the choice is free now and not later. nova records the name
it first saw and has no rename path, which is what makes this a
pre-deployment decision rather than something to fix afterwards.

Assisted-by: Claude:claude-opus-5
Signed-off-by: Roger Luethi <luethi@osism.tech>
OSISM 11 refuses to bootstrap hosts whose names are inconsistent, and
the release note describes that refusal. It reaches the wrong reader:
someone planning a deployment is in the configuration guide, choosing
the very names the check will judge, and the inventory page said
nothing about them.

State the rule where the decision is made. One form throughout, short
or fully qualified; the combination that breaks is fully qualified
inventory names without hostname_use_fqdn, where each machine answers
to one name while the network answers with another. Name the three
things that break, because "components may disagree" does not convey
that each one is fatal by itself.

List the other conditions the bootstrap checks -- length, case,
collisions, resolvability -- so the page describes what is enforced
rather than only the headline case. The collision example is two
fully qualified names differing after the first label, which is the
one an inventory can walk into without anything looking wrong.

Say that the choice is made before deploying, since nova stores the
name it first saw for a compute and cannot rename one.

Assisted-by: Claude:claude-opus-5
Signed-off-by: Roger Luethi <luethi@osism.tech>
@ideaship
ideaship marked this pull request as ready for review October 2, 2026 10:53

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: New

Development

Successfully merging this pull request may close these issues.

2 participants