Comment 1 for bug 210031

This may also be the cause of InstallFailed errors, as mentioned in OOPS-906SMPU119

<mwhudson> jml: every time i've look at an installfailed oops, the branch has been fine the next time
<jml> mwhudson: as in, the db records a successful mirror?
<mwhudson> yes
<mwhudson> by the time the oops report comes around even
<mwhudson> i've not been systematic about it though
<jml> in that case, I reckon that it's tied to the request_mirror race bug.