These are chat archives for rosshinkley/nightmare

19th
May 2016
James Tindall
@atomless
May 19 2016 09:36
@rosshinkley - that's correct, yes, it works on my local machine but not on the test automation machine - just trying to identify what differences there might be in the Jenkins test machine env - they have the same node/npm version and Nightmare.js version.
Ross Hinkley
@rosshinkley
May 19 2016 14:29
@atomless are they the same os? are both running "real" framebuffers? (I don't think a virtual framebuffer would cause a problem...)
if i have some time later today, i'll see if i can reproduce the problem
in the meantime, if you can put together a minimum repro case for your jenkins failure, that would be splendid
Ross Hinkley
@rosshinkley
May 19 2016 15:24
i marked a bunch of stuff for a minor release yesterday but got sidetracked
i will probably cut that release in an hour or so, barring calamity and anyone here protesting
Ross Hinkley
@rosshinkley
May 19 2016 17:19
mmmm... circle seems to not like this plan
Ross Hinkley
@rosshinkley
May 19 2016 17:30
it would appear circleci has changed how outbound requests are resolved in an effort to prevent denial of service attacks
Ross Hinkley
@rosshinkley
May 19 2016 17:48
i hopped into one of the temporary build boxes and poked around
the tests issuing requests against the bad domain are actually resolving :|
Rob Brackett
@Mr0grog
May 19 2016 17:49
o_O
Ross Hinkley
@rosshinkley
May 19 2016 17:49
yyyyeahhh
Rob Brackett
@Mr0grog
May 19 2016 17:49
sigh
Ross Hinkley
@rosshinkley
May 19 2016 17:49
i have a "fix"
gonna push it up as a pr in a sec
this may invalidate the tests, but instead of going to .com, i made it point to .tld
Rob Brackett
@Mr0grog
May 19 2016 17:50
does that work???
wtf
Ross Hinkley
@rosshinkley
May 19 2016 17:50
works insofar as the tests run and fail correctly, yes
Rob Brackett
@Mr0grog
May 19 2016 17:50
does Circle have any documentation about the change?
Ross Hinkley
@rosshinkley
May 19 2016 17:51
i've been poking around, and i can't find anything
mayhaps your googlefu is more powerful than mine
Rob Brackett
@Mr0grog
May 19 2016 17:51
unlikely
Ross Hinkley
@rosshinkley
May 19 2016 17:51
i'll also point out it's entirely possible that the "threat" was legit
although the merged tests should have happened in series, not in parallel
then again, i don't know what the threshold is for a dos
:P
"you issued more than three requests to the same domain in a minute! hacker!"
also bizarre: the PR tests appear to be passing
Rob Brackett
@Mr0grog
May 19 2016 17:53
oh, interesting, so it's rate limiting
Ross Hinkley
@rosshinkley
May 19 2016 17:53
eg the change you just pushed up for 640
Rob Brackett
@Mr0grog
May 19 2016 17:54
yeah
Ross Hinkley
@rosshinkley
May 19 2016 17:54
oh, i have no idea, i'm just guessing
Rob Brackett
@Mr0grog
May 19 2016 17:54
gonna guess it's because you merged all those PRs in quick succession, then
Ross Hinkley
@rosshinkley
May 19 2016 17:54
i'm letting circle cool off for a few minutes
then i'm going to nudge it again and see if it magically passes
if that's the case, then the threshold is ridiculously low
6 prs + 1 changeset x3 requsets -> 21 requests in ~10 minutes is too high?
:|
Rob Brackett
@Mr0grog
May 19 2016 17:56
¯\_(ツ)_/¯;
Ross Hinkley
@rosshinkley
May 19 2016 17:57
arright, it's been roughly 20 minutes
rosshinkley @rosshinkley cautiously clicks "rebuild"
Ross Hinkley
@rosshinkley
May 19 2016 17:59
:(
:angry:
Ross Hinkley
@rosshinkley
May 19 2016 18:05
segmentio/nightmare#649
Ross Hinkley
@rosshinkley
May 19 2016 18:13
huzzah, tests pass
Rob Brackett
@Mr0grog
May 19 2016 18:13
yay
Ross Hinkley
@rosshinkley
May 19 2016 18:13
yet another bulletpoint for my "irritation-driven development" seminar :P
okie dokie. nightmare 2.4.1 is out in the universe
Rob Brackett
@Mr0grog
May 19 2016 18:14
virtual mosquitos
Ross Hinkley
@rosshinkley
May 19 2016 18:15
virtual mosquitos... that's supervillain-caliber terrible
Maxwell Zhao
@thepiercingarrow
May 19 2016 22:51
@rickmed K, thanks anyway :)