Hi All,
Earlier today we had our weekly FxA DevOps catchup, and it wasn't
recorded, but the highlights were:
* Train 78 has been deployed, although we didn't manage to resolve one
particular (confidential) issue in time for this train and we'll need
to revisit it for train-79.
* Speaking of which, train-79 will be cut this week, and seems pretty
quiet from a deployment perspective.
* We discussed a bit whether it's useful to have a distinction between
NODE_ENV=stage and NODE_ENV=production, or whether it was a footgun.
The consensus seemed to be that it was a footgun, and we should use
individual config flags for things that we need to differ between
stage and prod, rather than doing an `if (NODE_ENV === 'production')`
test.
* We discussed this old PR about returning an error code when email
delivery fails, and decided it's worth picking up again as part of
the broader work on email deliverability:
https://github.com/mozilla/fxa-auth-server/pull/1329
* The marionette-based functional tests leak memory like a rusty sieve
with a hole burned through the middle; Vlad had some ideas that might
help though, such as enabling e10s.
* We discussed the logistics of deploying from a private repo, and how
to ensure that the deploy stays private all the way through to
production.
* We chatted a little about whether we could run all services under a
single hostname, rather than having separate hostnames for each.
Something like:
https://accounts.firefox.com/api/auth/v1/blah/bah
In general folks seemed cautiously optimistic about this; it would not
add a huge amount of operational complexity and would have a variety
of small wins, such as working better with Tor browser and avoiding
CORS pre-flight requests. It would be even better if we ever move
to HTTP2.
Cheers,
Ryan
_______________________________________________
Dev-fxacct mailing list
[email protected]
https://mail.mozilla.org/listinfo/dev-fxacct
