Merge ~bryce/ubuntu/+source/dovecot:sru-lp1991564-jammy into ubuntu/+source/dovecot:ubuntu/jammy-devel
| Status: | Merged | ||||
|---|---|---|---|---|---|
| Approved by: | git-ubuntu bot | ||||
| Approved revision: | not available | ||||
| Merged at revision: | f5217e4fc8246f2d7d673769ab398f827ef1795c | ||||
| Proposed branch: | ~bryce/ubuntu/+source/dovecot:sru-lp1991564-jammy | ||||
| Merge into: | ubuntu/+source/dovecot:ubuntu/jammy-devel | ||||
| Diff against target: |
1095 lines (+1049/-0) 7 files modified
debian/changelog (+12/-0) debian/patches/remove-unnecessary-master_service_flag_use_ssl_settings.patch (+112/-0) debian/patches/remove-unused-master_service_is_ssl_module_loaded.patch (+62/-0) debian/patches/series (+5/-0) debian/patches/split-master_service_ssl_settings_to_iostream_set-to-client-server-functions.patch (+258/-0) debian/patches/split-off-master_service_ssl_server_settings.patch (+473/-0) debian/patches/use-ssl-server-settings-only-when-necessary.patch (+127/-0) |
||||
| Related bugs: |
|
| Reviewer | Review Type | Date Requested | Status |
|---|---|---|---|
| git-ubuntu bot | Approve | ||
| Athos Ribeiro (community) | Approve | ||
| Canonical Server Reporter | Pending | ||
| Canonical Server Core Reviewers | Pending | ||
| Canonical Server | Pending | ||
|
Review via email:
|
|||
Description of the change
Straightforward fix, but it depends on several refactoring patches. I had earlier attempted to pick out just the changes needed to keep the patch minimal, however that did not prove to solve the issue. I figure maybe it's safer from a risk standpoint (and less effort for me) to ship the five patches unmodified from upstream, where they've gone through batteries of testing, rather than take a 2nd shot at a minimal backport, with the potential of adding defects through coding mistakes. If the changes are too hefty, the SRU team can push back, and that'll give clue that the additional effort for minimizing the patch would indeed be necessary.
I'd appreciate also a peek at the SRU text in the bug.
PPA: https:/
Test results:
- dovecot/
+ ✅ dovecot on jammy for amd64 @ 16.02.23 06:33:43 Log️ 🗒️
+ ✅ dovecot on jammy for arm64 @ 16.02.23 06:38:27 Log️ 🗒️
+ ✅ dovecot on jammy for armhf @ 16.02.23 06:10:42 Log️ 🗒️
+ ❌ dovecot on jammy for i386 @ 16.02.23 06:20:03 Log️ 🗒️
• doveadm FAIL 🟥
• systemd FAIL 🟥
• command1 FAIL 🟥
• testmails FAIL 🟥
+ ✅ dovecot on jammy for ppc64el @ 16.02.23 06:21:00 Log️ 🗒️
+ ✅ dovecot on jammy for s390x @ 16.02.23 06:10:19 Log️ 🗒️
The i386 failures are present in focal and kinetic as well. I haven't investigated them yet but suspect they're not going to be considered blockers.
| Bryce Harrington (bryce) wrote | # |
There was an error fetching revisions from git servers. Please try again in a few minutes. If the problem persists, contact Launchpad support.
Hi Bryce, Thanks for working on this one :)
The packaging side of this MP LGTM.
It's also nice that the package has some DEP8 tests and runs the upstream unit test suite during build time to give us more confidence on the regressions issue.
As for the SRU template, there is one bit that got my attention:
> User testing in production can be of value, since the principle impact is the side effects from hitting the fatal error, rather than the error itself.
I wonder if this should be an opt-in "feature" for users, since jammy is an LTS release. This could be done by staging the SRU (block-proposed) and inviting users to test the package for a while (we should expect that at least the reporter should test the changes).
However, there will still be the "risk" that we need to push another change to the package (a high priority bug or CVE) soon after this patch set lands in -proposed and the patch ends up landing in the updates/security pocket right away. Forcing users to perform production tests to ensure we are not adding regressions here.
Either way, this should be something for the SRU team to evaluate.
LGTM!