tests fail about que HOT 9 CLOSED

This started happening at 7c27845, tests all run fine before that.

from que.

I can't reproduce it on my machine, either on master or that commit, and things seem to be fine on Travis on 2.0.0 (just the same old intermittent errors on rbx and jruby). I don't know why there would be forked processes running - I played around with forking a bit locally, but it never made it into master. I notice that all the errors relate to ActiveRecord - what version is in your Gemfile.lock? I'm using 4.0.0.

from que.

4.0.2. Gemfile.lock is here: https://gist.github.com/joevandyk/e5a7233fef94d072266b/raw/a5bbef1c89cf9c629b9847f0655356eb45ab5116/gistfile1.txt

I'll try to dig some more.

from que.

Nevermind about the fork.

When the process was hanging, I used ctrl-c to stop the process, but for some reason, the ruby process still sticks around, see https://gist.github.com/joevandyk/e88eced0ae2c4d8f94a1 for the transcript. (then when I tried running rake again, there'd be multiples of these left-over ruby processes.

from que.

I just updated to 4.0.2, still no issue on 2.0.0-p247 or 2.0.0-p353.

Yeah, I've seen that issue with Ctrl+C not killing test processes properly, I'm not sure why it happens. Usually it just results in advisory-lock lint warnings, nothing like this.

from que.

Weird. I can replicate it on two separate ubuntu 12.04 machines and two separate OSX machines.

from que.

That's very, very strange. I'm running Linux Mint 15, which is derived from Ubuntu 13.04. I'll try to reproduce this on 12.04 tomorrow.

from que.

Figured it out. I have the TZ environment variable on my machine set to "UTC" so that Time.now returns values in that time zone. I set it up that way forever ago to emulate Heroku, but it's doing something unexpected with ActiveRecord here. I'm not sure what the fix is yet, but deleting that environment variable lets me recreate the problem on my machine, so I can figure it out from here. Thanks!

from que.

Looks good!

from que.