Katello/Pulp-Community Integration meeting minutes

2024-09-18 1000-1030 GMT-5

Attendees: ggainey, sjha, mdellweg, iballou, markusb, quba42

Regrets:

Agenda:

  • Previous AIs:

  • Getting “Structured APT” ready for the initial merge: https://github.com/Katello/katello/pull/11058

    • PR is “feature complete”, but still being tweaked

    • Currently working on the suggestions by jeremylenz

    • ATIX has performed extensive testing of the “feature present but disabled” case (ensures the refactor bits don’t break anything):

      • Export/Import was tested
      • Internal robotello tests were run
      • Large syncs general usage
    • So long as the “disabled” case does not break anything we can merge, even it the feature itself is not perfect yet. Call it “tech preview”, "optional or just don’t tell people about it at first.

    • When does Katello next branch?

      • 2 months-ish from “now”
      • next pulp-upgrade will be one-month-ish from now
      • prob want this merged pre-pulp-upgrade
    • Proposal:

      1. Finish the suggestions by jeremylenz
      2. Have Bernhard or Markus perform a final code review and wait for their ack.
      3. Final chance for core Katello review/request for changes
      4. Merge.
      5. Keep improving the feature in small follow up PRs.
    • net: merge state above with “disabled”; improve “enabled” functionality with new (smaller) PRs

  • [markusb] default ContentView with limited Repo visibility or ContentView providing library-instance repos.

    • There is now a thread for this: Transparent Content View for Limiting Repository-Access in Katello - RFCs - TheForeman
    • This had been possible by limiting subscribed products on an ActivationKey-level (SCA removed that possibility).
    • Scenario: a registerd host should
      1. must have access to latest synced content
      2. must only see a subset of Repos
    • net: lots of discussion on associated foreman thread, incl a proposal - is katello happy for ATX to work on implementing this proposal?
      • AI: sjha would like to have a meetup w/ ATIX/katello/subscription-DE to flatten some issues
  • [quba42] Is the following (satellite) issue also planned for upstream? https://issues.redhat.com/browse/SAT-25398

    • Related question: Is red hat bugzilla being replaced?
      • Yes - Satellite is on Jira now
    • not currently high on the backlog to fix
    • needs an upstream-redmine for community-to-work-on
  • [markusb] Should creation of IncrementalCV with not-available content (e.g. rpm not in any repository) throw an error-message?

    • ATIX hit internally
    • katello: prob useful to have an error msg
  • [quba42] I think this can be closed: Feature #32514: Improve deb repo form regarding simple vs. structured publishing - Katello - Foreman

    • old ask for “Structured Apt”
  • [mdellweg] Inquiry RE foreman (?) PRs from ATIX not getting appropriate/timely attention?

    • possibly not an issue - PTO Happens :slight_smile:
  • [quba42] Is Katello planning to somehow apply indices separately from the pulpcore migration? Publish and Promote slow with 3.11/4.13 - #11 by gvde - Support - TheForeman

Action Items:

1 Like

2024-10-01 1000-1030 GMT-5

Attendees: @iballou forgot to record attendees but it was more or less the “usual crew”

Regrets: not recording attendees

Agenda:

  • Structured-APT PR is ready for a final review
    • Tested on an internal production system by quba42
    • Just requires small remaining changes
  • Prototyping transparent/rolling/simplified content views starting soon
  • New release of pulp-deb would be good before Katello upgrades
    • Pain-point: updating pulp-deb Pulp CLI compatability each time Katello packages pulp-deb is cumbersome.
      • Potential fix: loosen the boundaries for pulp-deb Pulp CLI compatability. This could mean breaking changes get in, but it also means packaging happens more smoothly and people can get to testing it sooner.
  • Previous AIs:

Action Items:

3 Likes

2024-11-12 1000-1030 GMT-5

Attendees: ggainey, sjha, hstct, quba42, mbucher

Regrets: iballou

Agenda:

Action Items:

2024-12-03 1000-1030 GMT-5

Attendees: ggainey, quba42, mbucher, dalley, pbrochado, sjha, hstct, qjames, mdellweg

Regrets:

Agenda:

Action Items:

1 Like

2025-01-07 1000-1030 GMT-5

Attendees: sjha ggainey mbucher quba42 vsedmik hstct

Regrets:

Agenda:

Action Items:

1 Like

2025-03-04 1000-1030 GMT-5

Attendees: ggainey, vsedmik, qjames, quba42, mbucher, hstct

Regrets: iballou

Agenda:

  • Previous AIs:
  • [m-bucher]: Katello; Org with trailing . (e.g. ACME Inc.) => {"base_path":["The provided base path contains forbidden characters."]}
    • container-related - ORG with trailing-dot
    • Pulp is complaining, should katello sanitize?
    • is this container-only? pulp-content-app?
    • atix asking their support to experiment w/ deb/rpm
    • qjames: katello may have already fixed in 4.15-container-push
    • mbucher: will try against nightly
    • this may be specifically a not-legal-container-path Thing
    • if pulp is being “overly restrictive” around base-path-legality, we’ll need an issue to get that addressed
  • [quba42] Still open: Handle URL params for AptRepoFiles by quba42 · Pull Request #3454 · candlepin/subscription-manager · GitHub
    • ignored for 5 months?
    • let’s try and get tthis reviewed/merged
    • ptoscano is prob a person to nag
  • [quba42] Need packaging help: Update rubygem-pulp_deb_client to 3.5.1 by quba42 · Pull Request #11793 · theforeman/foreman-packaging · GitHub
    (Known)Issue or already fixed?
    • skip for now - needs a complicated-packaging-dance to address
    • might happen as part of the ongoing katello-branch-and-rebase-to-new-pulp
  • [quba42]: rolling CV: Fixes #38048 - Add rolling content views by quba42 · Pull Request #11240 · Katello/katello · GitHub
    • Katello 4.16 has branched so we want to finalize ASAP.
    • Lots of review and rework activity happening.
    • Open issues:
      • Smart proxy sync not re-syncing rolling CVs in all cases (maybe fixed now)
      • Missing repo names for rolling CVs on the smart proxy view
      • Container push repos need to be disallowed
      • Why is ::Actions::Katello::ContentView::AddToEnvironment async during rolling CV creation?
      • Hammer PR
      • docs PR
    • work is in-progress, will report if/when/as there may be roadblocks
    • PR might get taken over by mbucher
  • Availabilty: quba42 on leave for 4 weeks starting next week, nag hstct for pulp_deb or mbucher for katello-deb issues
  • [quba42] Potential future rolling CV extension: “I don’t want to sync all of Library to smart proxy just to use my hand full of rolling CV repos”
    • Possibility 1: Call it something different for rolling CVs, e.g.: “RollingLibrary” or just “Rolling”.
      • I don’t want to introduce another new “special” name.
    • Possibility 2: Allow “promoting” rolling CVs to other LCENVs?
      • Easy enough to implement, but does not make a lot of semantic sense.
    • Possibility 3: Allow some kind of filtering by CV on the smart proxy sync?
      • Sounds hard and potentially disruptive.
    • Any other ideas?
    • qjames: let’s rule out 1 above
    • qjames: 2 “makes more sense” (imnsho)
    • vsedmik: autosync-after-promotion
    • vsedmik: option 2 - breaks rolling-content-view “concept”
    • vsedmik: what about org-scoping libraries?
    • quba42: customers prob won’t be happy w/ introduced complications of org-scoping
    • consensus: proposal-1 is Right Out. Discussion needs to continue
    • finish the existing PR, and then perhaps open the discussion to a wider audience

Action Items:

1 Like

2025-04-01 1000-1030 GMT-5

Attendees: mbucher, sjha, ggainey

Regrets:

Agenda:

  • Previous AIs:
  • We do not do “April Fools!” jokes in this meeting! :slight_smile:
  • pulp_deb and domains
  • [m-bucher] container-image naming in katello revisited
    • see last mtg’s discussion
    • katello’s container-name-regex more permissive than Pulp’s
    • Pulp’s regex is different than OCI’s? - why?
    • current katello-PR should fix immediate issue
      • it’s OK if katello is stricter-than Pulp’s regex

Action Items:

1 Like

2025-05-06 1000-1030 GMT-5

Attendees: quba42, mbucher, hstct, sjha, ggainey, vsedmik

Regrets:

Agenda:

Action Items:

  • quba42: open a UX issue RE “green task that didn’t actually DO anything is…confusing”
  • quba42: update rolling-content-view/environments issue w/ discussion about approach being taken
  • hstct: work w/ SUSE to fix their repo?
  • ggainey: extend mtg to remainder of 2025
  • ggainey to add minutes to Katello/Pulp-Community Integration meeting minutes
1 Like

2025-06-03 1000-1030 GMT-5

Attendees: pbrochado, manisha, sjha, ggainey, mbucher, quba42, vsedmik, hstct

Regrets:

Agenda:

Action Items:

1 Like

2025-07-01 1000-1030 GMT-5

Attendees: ggainey, mbucher, qjames, quba42

Regrets:

Agenda:

  • Previous AIs:
    • None
  • quba42: Update on “rolling environments”: https://github.com/Katello/katello/pull/11407
    • sync-to-proxy fix almost ready
    • needs more unit-tests to cover all the state-changes
    • quba42 will undraft and ping when he thinks it’s ready
    • katello pushing hard towards 4.18 dev-freeze - be aware
  • Q: ATIX customer issue w/ update-capsule-repository task “starts, task-finds-no-capsules, dynflow shows green and then ‘eventually’ foreman-task turns red”?!?
    • transient/not-reproduceable
    • have we seen this before? (alas, no)
    • no issue yet, just reported
  • katello-dev-server on a Mac VM (aarch64)
    • everything works except pulpcore, because RPMs only avail for x86_64
    • Pulp team doesn’t build RPMs, TheForeman Build Gang does
    • Can you point foreman/katello at an “external/containerized” Pulp?
    • discussion ensues

2025-08-05 1000-1030 GMT-5

Attendees: quba42, sjha, mbucher, vsedmik, hstct, ggainey, pbrochado

Regrets:

Agenda:

Action Items:

2 Likes

2025

2025-09-02 1000-1030 GMT-5

Attendees: pbrochad, Bernhard S, Jan B, Quirin, Samir, Tobias, Vladmir, ggainey

Regrets:

Agenda:

  • Previous AIs:
    • update to py3.11 as minimum (py3.9 EOLs in October)
    • core<3.100 (should be done by the time this mtg happens)
    • both accomplished via 1311
  • quba42: Thanks to everyone involved with rolling CVs! https://github.com/Katello/katello/pull/11407
  • quba42, iballou: Rolling CV tests for robottelo
  • quba42, mbucher: Small open PR: https://github.com/Katello/katello/pull/11481
  • quba42: Migrate all deb content to use structured APT for Katello 4.19?
    • Draft PR: https://github.com/Katello/katello/pull/11487
    • How sure can I be that “during upgrade rake tasks” will really be run in all cases?
    • Where can I add docs to recommend users enable structured APT BEFORE upgrading to Katello 4.19 (or whatever version this will land in?)
  • quba42: Incremental CV update (RHEL/rpm content with >100 repositories and at least one filter) => OOM death
    • 188G of RAM is not enough
    • Not exactly good practice but apparently there are “reasons to do things this way”
    • Incremental CV update to bring back a errata
    • Why should this be a performance killer?
      • Even if there are many repos presumably only one of them needs the errata packages/copy action?
      • Same for the filter? (Does the filter perhaps apply to every repo even if there are no relevant packages?)
    • dep-solve on or off?
      • because “on” will try to dep-solve against All The Things
      • the claim is “off”
    • what is the form of the exact filter being used?
      • how is the filter applied? (can be limited to single repo, but if not, is applied to all of them)
    • Any Thoughts?
      • check on depsolve flags
      • check on where filter is applied
      • who can ATIX talk to with questions on CV-Incr-Update path - sjha or iballou
  • quba42: HTTP proxy with SSL inspection. Host can reach the repo, Pulp fails with “certificate verify failed: unable to get local issuer certificate”. Worked with Katello 4.14/pulpcore 3.49, not working with Katello 4.16/pulpcore 3.63 (both Python 3.11). Any ideas?
    • can we add CA explicitly? will that help?
  • how are things going w/ upstream user balu?
1 Like