-
Notifications
You must be signed in to change notification settings - Fork 3k
Random failures when building, no npm-debug.log present #6789
Comments
Thanks. That looks nasty. My first guess is some sort of race condition between building the two dependencies (contextify and ws) that require Can you add |
Will do! |
Got another random failure. full output is here: https://gist.githubusercontent.com/nullstyle/3c66cb86454e4946cc2f/raw/output.txt |
Thanks -- will review this afternoon On Tue, Nov 25, 2014 at 1:14 PM, Scott Fleckenstein
|
Nothing definitive yet. I do notice that ws gets installed twice - npm verb linkBins ws@0.4.32 Is this necessary? Does the shrinkwrap say to do that? (This shouldn't be causing the problem, but hey, grasping at straws here...) |
The annoying thing about The answer is probably in that second build log, but it may also be something that's impossible to reconstruct by a normal human being. It seems like maybe it's something to do with an interaction between the shrinkwrap-driven install and native module builds. I don't have time to look at this right now, but if you two make any progress in getting to the root cause of this, let me know. |
So here's a concrete request: can you add Is your Jenkins install local or hosted? If local, can you set it up so that the directory sticks around long enough for you to play around in it? I'm curious if any of these commands succeed (after the
On the one hand, no hurry; but on the other hand, sooner is better --- both because this is a bad build situation for you and it probably indicates a race condition / unreported failure somewhere inside npm, and I'd like to find and fix that. |
I'm having a very similar problem when running It happens intermittently and only with packages that use node-gyp. If I had to guess I think it might have something to do with node-gyp fetching node from https://nodejs.org/dist/. But it is just a guess.
|
That's a testable theory. What happens if you put
before the |
Tried clue 1Instead of running clue 2When running just Jenky work-aroundbecuase I can't get this app to build at all without this
|
What I'm wondering about is whether there's a race when multiple |
Hmm, that would certainly line up with this. It definitely seems that the
|
To further complicate things, if I remove all modules but the ones that compile binaries (including some non node-gyp ones) the build completes successfully. I ran about 150 builds with a bash script like this. If I have the full package.json and run builds with the same bash script almost every build fails. A gist of each package.json is below node v0.11.14 2 package.json files (1 fails consistently ~90% fail rate, 1 passes consistently ~99% success rate):package.json |
Hey y'all, Sorry for the non-responsiveness, was away on a trip. I'll catch back up with this issue today. |
Anyone get a chance to take a dive into this issue? |
After updating to npm 2.1.12 the issue went away. I believe the issue was do to 756f3d4 |
That would be cool if true. @nullstyle, @mlehner616 , can you try updating to 2.1.12 and see if the problem still persists? you should be able to add
in your CI scripts. |
@theverything and I were working together on this and determined this
|
@mlehner616 Great to hear that it seems to be fixed. @nullstyle Since there are a number of people subscribed here, I'd like to close this one - when you get back to being able to check it out, if there are still problems, you can create a new issue. How's that sound? (If we don't hear back within a week, we'll close this one.) |
@smikes that sounds fine... we havent seen a failure recently (the repo that it was happening in often hasn't been pushed to much lately). Thanks! |
Cool! Closing as resolved (for some). If you do run into issues again, @nullstyle, file a new issue and we can pick up the thread again. |
When building from jenkins I often get non-deterministic failures; Click rebuild and it will succeed the next time around. No npm-debug.log ends up getting created, but a sample stdout of a failed run looks like this:
As per https://twitter.com/seldo/status/534888065186414592, WTFFFFFFF
:)
The text was updated successfully, but these errors were encountered: