Re: Heroku & Boot Times

2013-01-23 Thread Phil Hagelberg
Scott Parker writes: > Jeroen - thanks for the advice. Yeah, SNAPSHOTS in prod is a poor practice > anyway, this gives me a good incentive to find them and kill them. Yeah, doing any dependency resolution at process launch time is brutal in production; apart from boot time issues it introduces v

Re: Heroku & Boot Times

2013-01-23 Thread Sean Corfield
On Wed, Jan 23, 2013 at 9:42 AM, Scott Parker wrote: > Static files on boot... dang. When I was first investigating our slow boot > time I swear I checked Enlive, but another quick glance at the source > indicates it's almost certainly a contributor. Expletive! I will start > capturing some metric

Re: Heroku & Boot Times

2013-01-23 Thread Weber, Martin S
oglegroups.com<mailto:clojure@googlegroups.com>" mailto:clojure@googlegroups.com>> Date: Wed, 23 Jan 2013 12:19:06 -0500 To: "clojure@googlegroups.com<mailto:clojure@googlegroups.com>" mailto:clojure@googlegroups.com>> Subject: Re: Heroku & Boot Times Hi

Re: Heroku & Boot Times

2013-01-23 Thread Scott Parker
Jeroen - thanks for the advice. Yeah, SNAPSHOTS in prod is a poor practice anyway, this gives me a good incentive to find them and kill them. Static files on boot... dang. When I was first investigating our slow boot time I swear I checked Enlive, but another quick glance at the source indicates i

Re: Heroku & Boot Times

2013-01-23 Thread Jeroen van Dijk
Hi Scott, We had some issues as well. SNAPSHOTS are likely to be an issue because they are re-checked at least once a day. So if your app needs a restart this will be re-checked and might slow down boot time. We also had problems due to this in combination with failing maven mirrors. It is probabl

Heroku & Boot Times

2013-01-23 Thread Scott Parker
Anyone else running a production website with Clojure in Heroku and struggling with boot time problems? After digging through our logs from the past month, I've noticed it's not uncommon to have a dyno crashed for awhile because of boot time problems. It seems especially likely when dynos are cycli