Conversation
The support matrix already lists RHEL 10 + Podman 5, but the host-prep docs still described only RHEL 8/9 and the RHEL 9 CNI path, which fails on RHEL 10. Co-authored-by: Cursor <cursoragent@cursor.com>
Elastic Docs Style Checker (Vale)Summary: 10 warnings found
|
| File | Line | Rule | Message |
|---|---|---|---|
| deploy-manage/deploy/cloud-enterprise/configure-host-rhel.md | 51 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/configure-host-rhel.md | 135 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/configure-host-rhel.md | 150 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/configure-host-rhel.md | 153 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/configure-host-rhel.md | 431 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/ece-ipv6-aws-setup.md | 627 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/migrate-ece-to-podman-hosts.md | 81 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/migrate-ece-to-podman-hosts.md | 166 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/migrate-ece-to-podman-hosts.md | 181 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
| deploy-manage/deploy/cloud-enterprise/migrate-ece-to-podman-hosts.md | 184 | Elastic.Spelling | 'Netavark' is a possible misspelling. |
The Vale linter checks documentation changes against the Elastic Docs style guide. To use Vale locally or report issues, refer to Elastic style guide for Vale.
Leave the existing edit-in-place storage.conf step as it already was. Co-authored-by: Cursor <cursoragent@cursor.com>
| * Follow your internal guidelines to create a RHEL 8 (the version must be >= 8.5), RHEL 9, Rocky Linux 8, or Rocky Linux 9 server or VM in your environment. | ||
| * Follow your internal guidelines to create a RHEL 8 (the version must be >= 8.5), RHEL 9, RHEL 10, Rocky Linux 8, or Rocky Linux 9 server or VM in your environment. | ||
|
|
||
| {applies_to}`ece: ga 4.2` RHEL 10 is supported with Podman 5 only. Do not install Podman 4 on RHEL 10, and do not follow the RHEL 9 CNI networking steps on RHEL 10. |
There was a problem hiding this comment.
I'm proposing a small change on this line that you are adding. Considering how we have evolved from RHEL 8 to RHEL 10 and that we still support 2 major versions of Podman and 2 network backends, I think it's a good time to introduce a short summary that will feel visually simple, not only for RHEL 10.
| {applies_to}`ece: ga 4.2` RHEL 10 is supported with Podman 5 only. Do not install Podman 4 on RHEL 10, and do not follow the RHEL 9 CNI networking steps on RHEL 10. | |
| The following Podman versions and network backends are supported: | |
| * RHEL 8 and Rocky Linux 8: Podman 4 with CNI | |
| * RHEL 9: Podman 4 or Podman 5 with CNI | |
| * Rocky Linux 9: Podman 5 with CNI | |
| * {applies_to}`ece: ga 4.2` RHEL 10: Podman 5 with Netavark |
The previous, together with the important note where we tell the users to ensure they use a supported combination of OS and Podman based on the support matrix should be enough here.
| sudo dnf -y install containernetworking-plugins | ||
| ``` | ||
|
|
||
| Do not install `containernetworking-plugins` on RHEL 10. That package is not available, and ECE on RHEL 10 uses Podman's default Netavark backend instead of CNI. |
There was a problem hiding this comment.
If the package is not available on RHEL 10, and if the note says explicitly that it's only for RHEL 9 and Rocky Linux 9 (I suspect it's probably needed for RHEL 8 but that's a different story), then I'd say this sentence doesn't add much value here. I'd suggest to remove it.
| Do not install `containernetworking-plugins` on RHEL 10. That package is not available, and ECE on RHEL 10 uses Podman's default Netavark backend instead of CNI. |
Later in the network backend configuration we tell the readers to keep the default backend.
| On RHEL 9 and Rocky Linux 9, merge this setting with the existing `network_backend="cni"` configuration rather than creating a duplicate `[network]` section. | ||
|
|
||
| {applies_to}`ece: ga 4.2` On RHEL 10, do not add `network_backend`. Leave that setting unset so Podman continues to use Netavark. |
There was a problem hiding this comment.
I think we can remove these two paragraphs. network_backend shouldn't be mentioned here in the dual-stack section.
| On RHEL 9 and Rocky Linux 9, merge this setting with the existing `network_backend="cni"` configuration rather than creating a duplicate `[network]` section. | |
| {applies_to}`ece: ga 4.2` On RHEL 10, do not add `network_backend`. Leave that setting unset so Podman continues to use Netavark. |
The only message for any OS (RHEL 8,9 or 10) would be to not create a duplicate network section, and I'd say that's already covered in the step description.
Mentioning here again network_backend and then mentioning the RHEL 10 difference makes the reading flow unexpected in the dual-stack section.
| ``` | ||
|
|
||
| Then open `/etc/containers/containers.conf` and, in the `[network]` section, set `default_network`. If the file or section does not exist yet, create it. On RHEL 9 and Rocky Linux 9, merge this setting with the existing `network_backend="cni"` configuration rather than creating a duplicate `[network]` section: | ||
| Then open `/etc/containers/containers.conf` and, in the `[network]` section, set `default_network`. If the file or section does not exist yet, create it. On RHEL 9 and Rocky Linux 9, merge this setting with the existing `network_backend="cni"` configuration rather than creating a duplicate `[network]` section. {applies_to}`ece: ga 4.2` On RHEL 10, do not add `network_backend`. Leave that setting unset so Podman continues to use Netavark: |
There was a problem hiding this comment.
| Then open `/etc/containers/containers.conf` and, in the `[network]` section, set `default_network`. If the file or section does not exist yet, create it. On RHEL 9 and Rocky Linux 9, merge this setting with the existing `network_backend="cni"` configuration rather than creating a duplicate `[network]` section. {applies_to}`ece: ga 4.2` On RHEL 10, do not add `network_backend`. Leave that setting unset so Podman continues to use Netavark: | |
| Then open `/etc/containers/containers.conf` and, in the `[network]` section, set `default_network`. If the file or section does not exist yet, create it. Make sure not to create a duplicate `[network]` section if one already exists. |
As before, I think the best solution is to not talk about the network backend in this specific doc, as it's not necessary.
There was a problem hiding this comment.
Ideally we should probably include a bullet for RHEL 8 / Rocky Linux 8.
Originally this step said for RHEL 9 only because on RHEL 9 we had to update the default backend to CNI, while in RHEL 8 I think the default backend is CNI already.
Now that we are converting the step to "configure network backend" and mentioning RHEL 9 and RHEL 10 in different bullets I'd say it's worthy a bullet for RHEL 8 saying something like:
For RHEL 8, keep the default CNI backend.
Or even...
For RHEL 8, ensure that Podman uses the CNI network backend. Although CNI is the default, configure
network_backend="cni"explicitly to avoid depending on the existing host configuration.
That would be aligned with:
2 participants
Summary
ECE 4.2 already supports RHEL 10 + Podman 5 on the support matrix, but the host-prep docs still listed only RHEL 8/9. Following the RHEL 9 steps on RHEL 10 fails:
containernetworking-pluginsis not available, and ECE keeps Podman's default Netavark backend instead of switching to CNI.Updates:
<9restriction)network_backend=cniHost-prep differences are based on the ECE Ansible role RHEL 10 path (
tasks/base/RedHat-10).Generative AI disclosure
Tool(s) and model(s) used: Cursor (Grok)
(AI-authored via Cursor 🤖)