search

browserify / wzrd.in

browserify as a service.
http://wzrd.in
MIT License
637 stars 96 forks source link

wzrd.in unstable, particularly on master [was: heroku instructions] #40

Closed tmcw closed 10 years ago

tmcw commented 11 years ago

Right now it looks like wzrd.in is down and 500'ing on most modules. I'll look into how tricky it would be to document setup on heroku so that users can split up the load more evenly.

jfhbrook commented 11 years ago

Right now it looks like wzrd.in is down and 500'ing on most modules

Uh-oh! @maxogden any ideas?

I'll look into how tricky it would be to document setup on heroku so that users can split up the load more evenly.

Heroku instructions would be great! I've tried deploying it to nodejitsu before, but that didn't work out very well due to memory constraints clashing with the use of npm. This might be better now, though.

Keep in mind, because browserify-cdn uses leveldb, which means two things:

1) This only really scales vertically, since caching data isn't shared between instances 2) Unless you take special provisions, a deploy on a fresh system will be missing everything cached prior. There's code in the commits to handle a dump/restore situation, but I believe it all got removed when I decided that nodejitsu's memory constraints were a dealbreaker.

tmcw commented 11 years ago

Pull at #41

max-mapper commented 11 years ago

hmm I restarted it and it seems okay now. not sure what was up, CPU + memory usage seemed normal

max-mapper commented 11 years ago

also wzrd.in isn't on browserify-cdn master because I think everything https://github.com/jesusabdullah/browserify-cdn/commit/0a1268df8d2ccf2ecdc4143d72f1166252ee3df0 is unstable for production (I got weird NPM errors and browserify-cdn would crash + flap), maybe it just needs to get updated to work when NPM is down

jfhbrook commented 11 years ago

Are there any stack traces, crash messages, etc., I can use for triage?

jfhbrook commented 10 years ago

I'm going to close this issue for now. I eventually plan to run this on my own machines so I can get access to the logs, which will help with such issues.