launchpad-dev team mailing list archive

[Aaron Bentley <aaron@xxxxxxxxxxxxx>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Bjorn Tillenius wrote:
> Just to clarify. The reason I'm asking these question is that we should
> think carefully before adding another moving part.

This was presented as a fait accompli at the Epic.  It seems quite late
to be questioning it.

> Foundations team has
> to do the initial setup, which requires time from LOSAs. Then the Code
> team has to take what Foundations did and integrate it with their work.

AIUI
 - RabbitMQ is slated to be deployed across the whole of Canonical, not
   just Landscape and Launchpad.
 - The incremental costs of supporting it in Launchpad are much lower
   than the initial cost of implementing it, which has already been paid
   by Landscape.

> I'm a strong believer in kiko's rule of software engineering: It will
> never work the first time. So this means even more work for Foundations,
> Code, and LOSAs. Will all these teams have time to take on this kind of
> work, given their current priorities?
> 
> Also, this adds another thing for the LOSAs to run and maintain.

AIUI, this is not something additional.  It has already been deployed
for Landscape.

> What
> happens if there's something wrong with the RabbitMQ server (software or
> hardware).

For Jobs, as long as we also continue to poll, a RabbitMQ failure would
reduce service to its current levels, not cause an all-out failure.

> Do we run multiple servers to avoid Launchpad stop working if
> one server fails?

I think that's a a question for IS.

Aaron
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkrOWpUACgkQ0F+nu1YWqI0azgCfehTCaervwy3OXwsahk/HzKSo
lZkAn18+6oMJtjPYqMMdwzOzSEt9geUy
=YcS3
-----END PGP SIGNATURE-----