[Planetlab-devel] symetric peering

[Thierry Parmentelat]

Tony

I've checked slice creation both ways and everything looks good. As you 
say everything seems to work fine so far.

Thanks again for setting up the /peers/ area, it is most useful to me 
for understanding what is going on

The other way around, you may wish to take a quick look at our own logs.
In particular you can see that a few entries in your own database would 
not get cached over here because they do not satisfy the various 
validate_ functions , I assume because they are legacy entries or due to 
code changes over time:
- a few obviously broken keys.
- a couple other slices use the nm-controller instantiation method, that 
our code does not currently support
- the slice-naming constraints rejects a few of your slices - that 
basically prevents these experiments from extending on our nodes

==
The thing is, planet-lab.eu runs a myplc based on planetlab-4_0-rc3 
locally rebuilt on apr. 3, so we could use an upgrade
(this is also why I screwed up with this email uniqueness patch that was 
not on my end..)

I am unclear about this upgrading policy.
Initially I wanted to be cautious and to first run a vanilla-PlanetLab 
code - even though our changes are in production within another 
onelab-private deployment - but then I quickly had to go for the web 
pages exception

Now is probably the right time for us to start injecting our changes in 
this production site.
This will take me some time however, because of the migrations in the DB 
that I'll need to somehow perform manually.
I will be out of town until friday, and will undertake this upgrade then.

Will keep you posted of course

Thanks again -- Thierry