Closed cji closed 3 years ago
The described process sounds good, but it might be overkill. On release day, what time are releases usually cut? We typically aim to lift embargoes at 9am PT (negotiable), so I'm wondering whether we could simply merge a public PR right around that time, and start the release after?
EDIT: To clarify, I think we should have the release notes ready & approved before hand, so the public PR doesn't need to be reviewed.
We’ve gotten to be a bit follow the sun. We sometimes do start building on morning of release day much closer to the international date line (eg: Japan, China, Europe) than Pacific Time’s start of day.
-- Tim Pepper Orchestration Lead VMware Open Source Technology Center
From: Tim Allclair notifications@github.com Reply-To: kubernetes/release reply@reply.github.com Date: Thursday, June 11, 2020 at 4:42 PM To: kubernetes/release release@noreply.github.com Cc: Kubernetes Release Robot release-managers-private@kubernetes.io, Team mention team_mention@noreply.github.com Subject: Re: [kubernetes/release] Adding security details to release notes (#1354)
The described process sounds good, but it might be overkill. On release day, what time are releases usually cut? We typically aim to lift embargoes at 9am PT (negotiable), so I'm wondering whether we could simply merge a public PR right around that time, and start the release after?
You received this message because you are subscribed to the Google Groups "release-managers-private" group. To unsubscribe from this group and stop receiving emails from it, send an email to release-managers-private+unsubscribe@kubernetes.iomailto:release-managers-private+unsubscribe@kubernetes.io. To view this discussion on the web visit https://groups.google.com/a/kubernetes.io/d/msgid/release-managers-private/kubernetes/release/issues/1354/642984046%40github.comhttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fa%2Fkubernetes.io%2Fd%2Fmsgid%2Frelease-managers-private%2Fkubernetes%2Frelease%2Fissues%2F1354%2F642984046%2540github.com%3Futm_medium%3Demail%26utm_source%3Dfooter&data=02%7C01%7Ctpepper%40vmware.com%7C819bf8f2ecce475c269908d80e612653%7Cb39138ca3cee4b4aa4d6cd83d9dd62f0%7C0%7C0%7C637275157721600211&sdata=Inm7fMRCsZySQBxXKTVpAh6%2BlvY9A9KtGq4b07r%2BvvM%3D&reserved=0.
Another approach, easier to implement IMHO, could be to have the PSC publish the vulnerability info on a known location in k/security and we would modify the release notes tool to scan for that data before cutting a new patch version. That file could tie together commits that have been cherry picked and have them treated separately in the doc and would enable us to get a unified security release note on all branches, with links to every related commit.
To make the release notes process independent from the embargo lift time, those files could be published at any time encrypted in kms with a key shared with one the releng groups (say k8s-infra-release-editors
). That group would need to have the kms/decrypter role in the project, of course. This would keep the vulnerability hidden right until the release is cut and leave full control of the disclosure with the PSC, no PR reviews needed.
Alright, sounds like a plan. How about this design proposal:
- pull-requests:
- 12345
- 67890
release-note: This release note has been modified
cve: # optional
id: CVE-2019-1010260
description: Really bad SSRF in KCM
published: 2020-10-05
score: 5.2
issues: # optional
- 9876
- 9875
krel [release-notes,changelog]
and release-notes
to be able to consume these mappings during release notes generation (recursively scraping directories and files)I think we could solve multiple issues with the above mentioned proposal:
The first rough idea (without proposing a KEP) would be to add the minimal feature of being able to map notes from A to B. Then we could demo that at one of the next SIG Release meetings and discuss possible process changes around it.
I like this. It would be a lot easier than trying to do all of patch/test on a separate repo, or making late changes to mainline repo with explicit synchronization to not release without all needed changes.
If we do this, I think we should still edit the PR on mainline for visibility after release.
One unfortunately gap that would remain is not having the information directly in mainline git commit history. We can’t rewrite history there. I’d prefer a solid git history that is authoritative, without being tied to GitHub artifacts, but…I’m not sure how many people use git directly anymore. Much of Kubernetes is tightly coupled with GitHub-specific artifacts at this point.
-- Tim Pepper Orchestration Lead VMware Open Source Technology Center
From: Sascha Grunert notifications@github.com Reply-To: kubernetes/release reply@reply.github.com Date: Tuesday, June 16, 2020 at 8:07 AM To: kubernetes/release release@noreply.github.com Cc: Kubernetes Release Robot release-managers-private@kubernetes.io, Team mention team_mention@noreply.github.com Subject: Re: [kubernetes/release] Adding security details to release notes (#1354)
I think we could solve multiple issues with the above mentioned proposal:
The first rough idea (without proposing a KEP) would be to add the minimal feature of being able to map notes from A to B. Then we could demo that at one of the next SIG Release meetings and discuss possible process changes around it.
You received this message because you are subscribed to the Google Groups "release-managers-private" group. To unsubscribe from this group and stop receiving emails from it, send an email to release-managers-private+unsubscribe@kubernetes.iomailto:release-managers-private+unsubscribe@kubernetes.io. To view this discussion on the web visit https://groups.google.com/a/kubernetes.io/d/msgid/release-managers-private/kubernetes/release/issues/1354/644824607%40github.comhttps://nam04.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgroups.google.com%2Fa%2Fkubernetes.io%2Fd%2Fmsgid%2Frelease-managers-private%2Fkubernetes%2Frelease%2Fissues%2F1354%2F644824607%2540github.com%3Futm_medium%3Demail%26utm_source%3Dfooter&data=02%7C01%7Ctpepper%40vmware.com%7C1efb302024e7456a626508d81206f04d%7Cb39138ca3cee4b4aa4d6cd83d9dd62f0%7C0%7C0%7C637279168306663727&sdata=sp8j157a5fB7ujspcbHPt7nT6h8ZKqEJp7oR1sV%2BnJI%3D&reserved=0.
/assign @puerco
/assign @puerco
/priority important-soon
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale
.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close
.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta. /lifecycle stale
/remove-lifecycle stale
Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale
.
Stale issues rot after an additional 30d of inactivity and eventually close.
If this issue is safe to close now please do so with /close
.
Send feedback to sig-testing, kubernetes/test-infra and/or fejta. /lifecycle stale
What would you like to be added:
The Product Security Committee would like the ability to include details about security fixes in release notes just prior to release announcements going out.
Why is this needed:
There are situations where a vulnerability with a Medium or Low severity may be fixed semi-publicly (for example with a public PR but not mentioning the security implications). When the commit(s) are cherry-picked, they would not include any of the CVE/impact/other details in the release notes.
@justaugustus provided an example of how this could work:
/cc @kubernetes/release-engineering /cc @kubernetes/product-security-committee