1174073 - changeset propagation from the hg master to the webheads broke (larch repo)

Status: RESOLVED FIXED

(Developer Services :: Mercurial: hg.mozilla.org, defect)

Description

[Paul Rouget [:paul]]

We can't push to larch anymore:

> pushing to ssh://hg.mozilla.org/projects/larch
> searching for changes
> remote has heads on branch 'default' that are not known locally: 5b5a2fd267f1
> abort: push creates new remote head c131b9031158!
> (pull and merge or see "hg help push" for details about pushing new heads)

Not sure what happened. And I don't know how to solve this.

It's started here: https://bugzilla.mozilla.org/show_bug.cgi?id=1172422#c4

Comment 1

[Paul Rouget [:paul]]

Gps, maybe you can help?

Comment 2

[Gregory Szorc [:gps]]

> pushing to ssh://hg.mozilla.org/projects/larch
> searching for changes
> remote has heads on branch 'default' that are not known locally: 5b5a2fd267f1

This is saying the repo you are pushing to has content that you don't have locally.

> abort: push creates new remote head c131b9031158!

This is saying that your push would create divergence on the server. e.g. here's the current state:

<common base> --- <unknown commits> --- 5b5a2fd267f1 (remote head)
              \
               \ 
                - <unknown commits> --- c131b9031158 (local head)

> (pull and merge or see "hg help push" for details about pushing new heads)

This is generic advice from Mercurial. It also happens to be wrong. At Mozilla, we generally prefer linear repository histories as opposed to merging. So, the solution to this is:

$ hg pull <remote>
$ hg rebase -d tip

This will fetch content from the remote repo and then rebase the current head on top of what you just pulled.

You may also want to read https://mozilla-version-control-tools.readthedocs.org/en/latest/hgmozilla/bookmarks.html#understanding-head-based-development

Comment 3

[[:fabrice] Fabrice Desré]

The thing is that https://hg.mozilla.org/projects/larch/rev/5b5a2fd267f1 doesn't exist... This was created when I pushed in bug 1172422. I got no error on my side, but that was just when the troubles started that led to the super long tree closure.

Comment 4

[Paul Rouget [:paul]]

`hg rebase` tells me there's nothing to rebase. `hg merge` tells me there's nothing to merge.

Comment 5

[Paul Rouget [:paul]]

hg incoming from https and ssh gives me different results.

>  <Pike> then the changeset propagation from the hg master to the webheads broke

Comment 6

[Gregory Szorc [:gps]]

I triggered a sync from ssh master to the http mirrors and confirmed the ssh and http servers are in sync.

Comment 7

[u429623]

s/b good - reopen and/or ping in #vcs if you see any remaining issues