spurious oopses from resolve_lp_path

Bug #246591 reported by Curtis Hovey
12
Affects Status Importance Assigned to Milestone
Launchpad itself
Fix Released
Medium
Michael Hudson-Doyle

Bug Description

I've seen resolve_lp_path occur several times over the past few days, which is always reported as no such project. The most frequent group of incidents is in https://devpad.canonical.com/~matsubara/oops-summaries/edge-2008-07-07.html

InvalidBranchIdentifier: <Fault 190: "Invalid branch identifier: 'bzr-svn/0.4/0.4.10'"> OOPS-927EB97

InvalidProductIdentifier: <Fault 240: '~dmindebian cannot be the name for a project.'> OOPS-927ED128

NoBranchForSeries: <Fault 170: u'Series trunk on software-properties has no branch associated with it'> OOPS-927ED121

NoSuchPersonWithName: <Fault 200: 'No such person or team: '> OOPS-927EC109

NoSuchProduct: <Fault 10: 'No such project: trunk'> OOPS-927EA25

NoSuchSeries: <Fault 180: u'Project bzr-svn has no series called "0.4.10"'> OOPS-927EB98

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

Oh grr. I guess there was a reason the old code was written the way it was then...

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

There are two fixes, really: go back to having the endpoint return, rather than raise, Faults or have the publication machinery not regard Faults being raised from xmlrpc stuff as something to log an oops about. I don't know which would be better.

Changed in launchpad:
importance: Undecided → Low
status: New → Confirmed
description: updated
Revision history for this message
Diogo Matsubara (matsubara) wrote :

Increasing importance. Not high or critical because users already get a nice error message when such Faults are raised. Medium because we don't want to clog the oops summaries with things that are not issues.

Changed in launchpad:
importance: Low → Medium
Revision history for this message
James Henstridge (jamesh) wrote :

The documentation in lib/zope/app/publisher/xmlrpc/README.txt only gives an example of returning a Fault object, and that is what all of our other code was doing. I'd recommend continuing to do so in resolve_lp_path().

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

Well, I think that's really stupid :) But it's also easy enough to change to.

Changed in launchpad:
assignee: nobody → mwhudson
status: Confirmed → In Progress
Revision history for this message
Diogo Matsubara (matsubara) wrote :

Hi Michael, how's this fix coming? Did you reach an agreement with jamesh?

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

revision 6832 of RF launchpad/devel fixes this -- sorry for the delay, got hit by sprints!

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

Rolled out with 2.1.8, but as this only really affected the edge server, it's been 'released' for a while!

Curtis Hovey (sinzui)
visibility: private → public
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.