RevisionNotPresent error when mirroring stacked branches

Bug #280595 reported by Jonathan Lange
2
Affects Status Importance Assigned to Milestone
Bazaar
Fix Released
High
Martin Pool
Launchpad itself
Invalid
High
Unassigned

Bug Description

We appear to be getting RevisionNotPresent errors when we mirror hosted stacked-on branches.

See oopses OOPS-1012SMPU11 and OOPS-1012SMPU18 for examples.

The user impact of this bug is unclear, since the branches all appear functional by the time the OOPS reports are sent out. This would appear to point to some sort of race condition where the first attempt to mirror a branch fails.

Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

I'm sure this is where the puller tries to mirror the branch before stacking is set up. Not sure what we can do about it, beat up the bazaar developers so the branch is only unlocked once?

Revision history for this message
Jonathan Lange (jml) wrote : Re: [Bug 280595] Re: RevisionNotPresent error when mirroring stacked branches

On Fri, Oct 10, 2008 at 1:26 PM, Michael Hudson
<email address hidden> wrote:
> I'm sure this is where the puller tries to mirror the branch before
> stacking is set up. Not sure what we can do about it, beat up the
> bazaar developers so the branch is only unlocked once?

We should do *something* about it. The OOPS reports are a pain.

jml

Revision history for this message
Stuart Bishop (stub) wrote :

I have a branch (lp:~stub/launchpad/pending-db-patches) that when I push I get an exception 'Must end write group before releasing write lock...' (Bug #230902). This exception seems to be masking the real exception - ERROR: Revision {<email address hidden>} not present in "patch121090.sql-20080212094058-j5r4p768zs6h6thb-1"

Would these branches be the trigger?

Revision history for this message
Michael Hudson-Doyle (mwhudson) wrote :

Stuart: no, that's something else -- branches failing to autopack properly, we think.

Revision history for this message
Leonard Richardson (leonardr) wrote :

I get this error: see OOPS-1022SMPU11.

Aaron Bentley (abentley)
Changed in launchpad-bazaar:
importance: Undecided → High
Aaron Bentley (abentley)
Changed in bzr:
assignee: nobody → abentley
Jonathan Lange (jml)
Changed in launchpad-bazaar:
status: New → Triaged
Changed in bzr:
importance: Undecided → High
status: New → Triaged
Revision history for this message
Aaron Bentley (abentley) wrote :

Martin, I believe your pending patch fixes this also

Changed in bzr:
assignee: abentley → mbp
Revision history for this message
Jonathan Lange (jml) wrote :

So, as best as we can tell, the Launchpad bug was occurring because Bazaar was making broken branches (see bug 291046). This bug has been fixed in bzr.dev and should be in Bazaar 1.11, which means that we can expect the number of branches created this way to go down.

Although it's conceivable that we could do something server side to fix existing broken branches, and also to fix incoming broken branches, I don't think we want to spend that much effort working around the GIGO rule.

Changed in launchpad-bazaar:
status: Triaged → Invalid
Martin Pool (mbp)
Changed in bzr:
status: Triaged → Confirmed
Jelmer Vernooij (jelmer)
tags: added: stacking
Revision history for this message
Martin Pool (mbp) wrote :

We no longer seem to be getting these problems, and indeed the mirrored codehost area is gone. Of the back of Aaron's comment #6 I'm going to call this closed.

Changed in bzr:
status: Confirmed → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.