Launchpad push with insane network download

[Martin (gzlist)]

On 20/04/2010, Robert Collins <robertc at robertcollins.net> wrote:
> Some highlights:
>
> 7% in suspend write group
> 24% in check_new_inventories
> 25% compressing (compress + flush)
> 20% reading content
>
> Not sure what to infer from this yet.

Is this along the same lines as bug 562666?

In which case, we're back to the two initial (possibly launchpad
specific) issues:
* The server hang on first push (perhaps from that kind of
cpu-expensive operation)
* The incomplete server state triggering a client whole-repo download

I think I'm out of ideas for experiments this end.

Martin