git-review doesn't warn about duplicate change-id
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Core Infrastructure |
Invalid
|
Undecided
|
Jeremy Stanley | ||
git-review |
Confirmed
|
Wishlist
|
Unassigned |
Bug Description
I'm having a branch with 9 patches, but when sending or updating it over Gerrit, it only sends the 6 first:
% git review
You have more than one commit that you are about to submit.
The outstanding commits are:
805699c (HEAD, jd/use-
e173528 service: run common initialization stuff
f197c10 Use console scripts for ceilometer-api
7183fd5 Use console scripts for ceilometer-dbsync
d0a5d09 Use console scripts for ceilometer-
daa393b Use console scripts for ceilometer-
bab2ca7 agent-central: use CONF.import_opt rather than import
b2d7ecc Move os_* options into a group
522e632 Use console scripts for ceilometer-
Is this really what you meant to do?
Type 'yes' to confirm: yes
remote: Resolving deltas: 100% (114/114)
remote: Processing changes: updated: 6, refs: 1, done
To ssh://<email address hidden>
* [new branch] HEAD -> refs/publish/
Check the number of updated patches: 6!
The last 3 are not marked as sent, I need to rerun git review:
% git review
You have more than one commit that you are about to submit.
The outstanding commits are:
805699c (HEAD, jd/use-
e173528 service: run common initialization stuff
f197c10 Use console scripts for ceilometer-api
7183fd5 Use console scripts for ceilometer-dbsync
d0a5d09 Use console scripts for ceilometer-
daa393b Use console scripts for ceilometer-
bab2ca7 agent-central: use CONF.import_opt rather than import
b2d7ecc Move os_* options into a group
522e632 Use console scripts for ceilometer-
Is this really what you meant to do?
Type 'yes' to confirm: yes
remote: Resolving deltas: 100% (55/55)
remote: Processing changes: updated: 3, refs: 1, done
To ssh://<email address hidden>
* [new branch] HEAD -> refs/publish/
And now the final 3 are finally marked as updated.
However, there's still a bug since the patch #7 "Use console scripts for ceilometer-api" never shows up in Gerrit. Worst, patch #8 shows up at https:/
So rather than having a patch tree like:
#1 -> #2 -> #3 -> #4 -> #5 -> #6 -> #7 -> #8 -> #9
built in one upload, I need to run twice git review and I end up with:
#1 -> #2 -> #3 -> #4 -> #5 -> #6
#8 -> #9
Changed in openstack-ci: | |
status: | New → Triaged |
assignee: | nobody → Jeremy Stanley (fungi) |
I'm suspecting something is/was weird about change #7 in the series. We've seen similar issues when a patch makes no substantive changes, or duplicates another patch's change-id header or sometimes becomes an orphaned commit because its parent gitsha doesn't match the one already present in Gerrit. Can you push a copy of the problem branch somewhere public so I can look at the change which didn't make it into a review?