Instant global deployments
First attempt with v2 and failuresNovember 13, 2018 at 3:33pm
I apologise Team Zeit, as I'm a magnet for bugs.
I've just uploaded my very first attempt at v2 release, a simplistic project that I use for my workshops. It took about 5 minutes and it failed.
This is the output: https://gist.github.com/remy/c0569803ce98bf9b3328dc6313f999a2
The logs are under https://next-ffconf2018-workshop-ljk40wl65.now.sh/_logs
It's complaining that it's too large but I can't see how.
Is there anything else I can provide to help debug? (I'm running it again, but already it's taking a long time - way way longer than with the v1 instance API).
November 13, 2018 at 4:43pm
@remy I can't see your logs. They've been made private by default now on v2, so you probably just need to set
But yes, those lambdas look way too big
Post a new link with public logs and I'll have a look
I'm uploading two public versions - on v1 + v2 I'll post links once it's "done" (or gives up)
dynamic - with server backend.
currently doing that hanging thing…
(damn, spectrum push notifications are annoying - keep sliding in when I'm on the page, then clicking clears the page and I have scroll down again!)
looks like it's stuck in a loop (sorry - currently treating this spectrum forum like a chat forum…)
So, here's the v2 deploy: https://zeit.co/rem/next-ffconf2018-workshop/k7bqj0dcf - completely hosed, all I did was to add v2 and the build option in the [now.json](https://zeit.co/rem/next-ffconf2018-workshop/k7bqj0dcf/source?f=src/now.json) (as per copy/paste docs).
And here's the v1 deploy: https://remyffconf-workshop-fuelaxqukw.now.sh/ - same code base, just no version or build in now.json - works.
It's pretty realtime, I've been treating it more like chat as well
Nice on the MDX. One unrelated note already, is that you may want to lazy load
next.config.js, so that you can move it to
Next gives you phase constants, which you can use for a conditional for dev mode
it's only loaded in that initial boot though, right? So it's not a huge concern (but definitely a better practise). (The codebase is for my updated workshop)
But I don't see why it fails so horrendously