Sort certificates in bundles to ensure deterministic behaviour

[jabdoa2]

fix #310

[cert-manager-prow]

Hi @jabdoa2. Thanks for your PR.

I'm waiting for a cert-manager member to verify that this patch is reasonable to test. If it is, they should reply with /ok-to-test on its own line. Until that is done, I will not automatically test new commits in this PR, but the usual testing commands by org members will still work. Regular contributors should join the org to skip this step.

Once the patch is verified, the new status will be reflected by the ok-to-test label.

I understand the commands that are listed here.

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.

[erikgb]

/ok-to-test

[jabdoa2]

Should be good to go now. I adjusted the certificate order in smoke and integration test.

[jabdoa2]

@erikgb is there a chance to get this merged & released soonith? we would prefer to forward fix our issue instead of rolling everything back.

[erikgb]

@erikgb is there a chance to get this merged & released soonith? we would prefer to forward fix our issue instead of rolling everything back.

@jabdoa2 Yes, it should be possible to merge and release this improvement soonish. I am going to review it today, but I think we should get #375 merged first. The other PR changes overlapping area of code to this PR, and has been open for a while. So you'd have to be prepared for a rebase. 😉

[erikgb]

Thanks for looking into this, @jabdoa2! Looks (almost) good to me. 😉 As already commented, we should probably wait for #375 to get merged.

[arsenalzp]

Thanks for looking into this, @jabdoa2! Looks (almost) good to me. 😉 As already commented, we should probably wait for #375 to get merged.

You can merge it as it is then I can re-write code to remediate all remarks as well as introducing crypto/x509 CertPool I'm on vacation right now, thus no chance to get laptop until Tuesday. I would like to stay the course.

[erikgb]

You can merge it as it is then I can re-write code to remediate all remarks as well as introducing crypto/x509 CertPool I'm on vacation right now, thus no chance to get laptop until Tuesday. I would like to stay the course.

Thanks @arsenalzp!

[arsenalzp]

You can merge it as it is then I can re-write code to remediate all remarks as well as introducing crypto/x509 CertPool I'm on vacation right now, thus no chance to get laptop until Tuesday. I would like to stay the course.

Thanks @arsenalzp!

How it will look like: my PR is the first one then #380 is the second one?
Or just merge #380 then I will re-write my PR?

[erikgb]

How it will look like: my PR is the first one then #380 is the second one? Or just merge #380 then I will re-write me PR?

I think #380 (this PR) can be merged first, and it also fixes a serious issue. It looks almost GTM to me.

[erikgb]

/lgtm

We should maybe squash commits before/on merge?

[jabdoa2]

/lgtm

We should maybe squash commits before/on merge?

I squashed everything into one commit

[erikgb]

Thanks for fixing this serious issue, @jabdoa2! 🚀

/lgtm
/approve

[cert-manager-prow]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: erikgb

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• OWNERS [erikgb]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[maelvls]

I spent Friday afternoon manually testing the bug with label selectors. I was able to reproduce it, and I was able to test that your PR fixes the issue. I have documented the testing I did in the Hackmd page PR 380: Manually Testing that ConfigMaps are No Longer Unexpectedly Updated By The ConfigMap Label Selector Bug.

By the way, would there be by any chance someone relying the fact that the ordering in sources is identical in targets?

[erikgb]

By the way, would there be by any chance someone relying the fact that the ordering in sources is identical in targets?

Order doesn't matter here IMO. The order must be consistent, but I don't think a particular order is required.

[jabdoa2]

By the way, would there be by any chance someone relying the fact that the ordering in sources is identical in targets?

Order doesn't matter here IMO. The order must be consistent, but I don't think a particular order is required.

This should be very unlikely for trust bundles. Those are not usually ordered and order should not matter.

The story would be different for PEMs containing a certificate chain. For instance, OpenSSL is well known for behaving badly if the first cert is not the leaf cert. Most other libraries will behave well though. Also that is not the usecase for trust-manager.