[Pyogp] How often and when to run Interop tests against Second Life Aditi

[Tess Chu]

We met after the Open Grid team huddle today to decide how often we
should run the pyogp interoperability tests against our code. We
decided that they should be run at the following frequencies:

* with every checkin for code changes
* after every deploy to aditi
* once a day at approximately 10am or 11am to coordinate with typical
deploys to Aditi

This will ensure that we catch changes to the system as soon as they happen.

Cheers,
Tess

[Enus Linden]

I'll get this set up asap, Thursday seems reasonable, given my schedule.

Prospero, there should be no infrastructure requirements. I'll just
monitor deploys closely for the foreseeable future.

Question about the first trigger, checkins... how can one monitor this,
and why would a checkin need to kick off the interop tests? Does it
imply code running on the agent domain has changed?

Was there discussion on the best way to monitor or publish the results
of the test runs?

enus

Post by Tess Chu
We met after the Open Grid team huddle today to decide how often we
should run the pyogp interoperability tests against our code. We
* with every checkin for code changes
* after every deploy to aditi
* once a day at approximately 10am or 11am to coordinate with typical
deploys to Aditi
This will ensure that we catch changes to the system as soon as they happen.
Cheers,
Tess

[Christian Scholz]

Hi!

Good news to have such tests in place! :)
But maybe you should have a stable version of these interop tests
running so that not the tests themselves break at some point because
somebody introduced a bug ;-)

Post by Enus Linden
I'll get this set up asap, Thursday seems reasonable, given my schedule.
Prospero, there should be no infrastructure requirements. I'll just
monitor deploys closely for the foreseeable future.
Question about the first trigger, checkins... how can one monitor this,
and why would a checkin need to kick off the interop tests? Does it
imply code running on the agent domain has changed?

As this only affects LL code and checkins I still wonder if you are
planning on running those tests as some pre-commit hook. I am wondering
then how long that might take and if people wouldn't stop checking in
bits of code.

And maybe they should run on some test grid before a deployment? As for
us outside here Aditi is sort of the live system.

Post by Enus Linden
Was there discussion on the best way to monitor or publish the results
of the test runs?

It would definitely be good to know what the state of aditi is also from
the outside. Easier to know then if it's a bug in my code or in your
code ;-)

-- Christian

[Tess Chu]

Good point about checkins not really affecting what's deployed, thus
reduces the need to run the tests. Lots of time we deploy changes we
made by hand, which would fall more into the deploy category.

Regarding publishing and monitoring, we were going to talk to Cg to see
if he has any recommendations.

Tess

Post by Enus Linden
I'll get this set up asap, Thursday seems reasonable, given my schedule.
Prospero, there should be no infrastructure requirements. I'll just
monitor deploys closely for the foreseeable future.
Question about the first trigger, checkins... how can one monitor
this, and why would a checkin need to kick off the interop tests? Does
it imply code running on the agent domain has changed?
Was there discussion on the best way to monitor or publish the results
of the test runs?
enus

Post by Tess Chu
We met after the Open Grid team huddle today to decide how often we
should run the pyogp interoperability tests against our code. We
* with every checkin for code changes
* after every deploy to aditi
* once a day at approximately 10am or 11am to coordinate with typical
deploys to Aditi
This will ensure that we catch changes to the system as soon as they happen.
Cheers,
Tess