K-C9 · size M · part of inbuxa for Kubernetes (tracking issue in inbuxa-charts)
Refuse a 2026.x to 27.1 crossing without upgrade.acknowledgeLayout; optional snapshot before upgrade; pre-upgrade Job running --upgrade-store; scale server to 0 first if S5 says so.
Done when
Lab upgrade tests green (K-L5)
Depends on: K-S3, S5
**K-C9** · size M · part of inbuxa for Kubernetes (tracking issue in inbuxa-charts)
Refuse a 2026.x to 27.1 crossing without upgrade.acknowledgeLayout; optional snapshot before upgrade; pre-upgrade Job running --upgrade-store; scale server to 0 first if S5 says so.
**Done when**
- Lab upgrade tests green (K-L5)
**Depends on:** K-S3, S5
jcoffey-dev
added this to the 27.1 milestone 2026-11-18 05:14:44 +00:00
Built and merged on 11-18. This stays open on the tracker until K-L5 is green.
inbuxa/inbuxa-server#457: inbuxa --upgrade-plan reports the layout steps an upgrade would run and whether one needs every node stopped. It changes nothing.
#17: server.upgrade, a pre-upgrade/pre-rollback Job. It refuses a layout change until acknowledgeLayout names the target version. When a step isn't rolling-safe, it scales the server Deployments to 0 first. Then it runs --upgrade-store. The scaling rights exist only while the hook runs.
inbuxa/inbuxa-k8s-lab#18: lab/chart/upgrade-test.sh. All three cases pass on kind (no change; refused; acknowledged 2→3 in 189 s with mail working afterwards). The crossing there is simulated by setting the store's mark back to 2 on the same image.
Not done yet:
K-L5 with real images: rc.N-1 → rc.N with mail flowing, and a 2026.x store with data crossing to 27.1.
The optional snapshot before the upgrade.
Docs (K-D6): helm upgrade --timeout 20m.
Built and merged on 11-18. This stays open on the tracker until K-L5 is green.
- inbuxa/inbuxa-server#457: `inbuxa --upgrade-plan` reports the layout steps an upgrade would run and whether one needs every node stopped. It changes nothing.
- #17: `server.upgrade`, a pre-upgrade/pre-rollback Job. It refuses a layout change until `acknowledgeLayout` names the target version. When a step isn't rolling-safe, it scales the server Deployments to 0 first. Then it runs `--upgrade-store`. The scaling rights exist only while the hook runs.
- inbuxa/inbuxa-k8s-lab#18: `lab/chart/upgrade-test.sh`. All three cases pass on kind (no change; refused; acknowledged 2→3 in 189 s with mail working afterwards). The crossing there is simulated by setting the store's mark back to 2 on the same image.
Not done yet:
- K-L5 with real images: rc.N-1 → rc.N with mail flowing, and a 2026.x store with data crossing to 27.1.
- The optional snapshot before the upgrade.
- Docs (K-D6): `helm upgrade --timeout 20m`.
K-L5 ran green with real images (inbuxa/inbuxa-k8s-lab#30): rc.33 → rc.34 rolled through the hook with mail flowing, and a real 2026.11.18.3 store crossed to rc.34 once acknowledged, mail intact. That meets the done-when here except the optional pre-upgrade snapshot, which isn't built.
K-L5 ran green with real images (inbuxa/inbuxa-k8s-lab#30): rc.33 → rc.34 rolled through the hook with mail flowing, and a real 2026.11.18.3 store crossed to rc.34 once acknowledged, mail intact. That meets the done-when here except the optional pre-upgrade snapshot, which isn't built.
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
K-C9 · size M · part of inbuxa for Kubernetes (tracking issue in inbuxa-charts)
Refuse a 2026.x to 27.1 crossing without upgrade.acknowledgeLayout; optional snapshot before upgrade; pre-upgrade Job running --upgrade-store; scale server to 0 first if S5 says so.
Done when
Depends on: K-S3, S5
Folded into the checklist on inbuxa/inbuxa-charts#14 to keep the issue list short. The details stay here; reopen this when work on it starts.
Built and merged on 11-18. This stays open on the tracker until K-L5 is green.
inbuxa --upgrade-planreports the layout steps an upgrade would run and whether one needs every node stopped. It changes nothing.server.upgrade, a pre-upgrade/pre-rollback Job. It refuses a layout change untilacknowledgeLayoutnames the target version. When a step isn't rolling-safe, it scales the server Deployments to 0 first. Then it runs--upgrade-store. The scaling rights exist only while the hook runs.lab/chart/upgrade-test.sh. All three cases pass on kind (no change; refused; acknowledged 2→3 in 189 s with mail working afterwards). The crossing there is simulated by setting the store's mark back to 2 on the same image.Not done yet:
helm upgrade --timeout 20m.K-L5 ran green with real images (inbuxa/inbuxa-k8s-lab#30): rc.33 → rc.34 rolled through the hook with mail flowing, and a real 2026.11.18.3 store crossed to rc.34 once acknowledged, mail intact. That meets the done-when here except the optional pre-upgrade snapshot, which isn't built.
Done: inbuxa/inbuxa-k8s-lab#30 merged.