"Socket hang up" Errors #311

blakmatrix opened this Issue Sep 14, 2012 · 4 comments

2 participants


"Socket hang up" Errors are vague and frustrating to see. Please add a better copy to better describe what is happening.


Users will see this when


|# | description | identified |
|1 | client times out | yes |
|2 | the server runs out of memory on a npm install (sometimes) | no |
|3 | The haibu on the drone is not running | no |
|4 | If there are zombies on the drone that need to be cleaned up | no |
|5 | If during the deploy the drone (related to client timeing out) takes to long to do the npm install (doesn't ENOMEM but just stalls) | ~maybe~ |
|6 | if the haibu is in a crash loop on the drone | no |

IS there a suggestion as to what the error should say in other cases? I'm not too sure we can differentiate most of them, but I am fairly certain we can determine client vs. server hang ups


@blakmatrix I want you to pursue this. I feel you can both identify these sorts of errors and come up with better copy. If there's anything weird, we'll catch it in your PR.


interesting case... https://gist.github.com/97e39da7ca611bf7a4df the app seemed to crash on the drone... but a second deploy with same code deployed just fine


Also discovered one that can be identified

       if(err.message && err.message === 'socket hang up'){
          if (err.code){
            if (err.code === 'ECONNRESET'){
              // MESSAGE explaining that the client gave up waiting for a response
              // and that it is possible that users code may still have deployed

I got this to happen by deploying a large framework... bypassing the build server using the devDependencies bug, effectively attempting to compile on the drone itself

@blakmatrix blakmatrix closed this Oct 2, 2012
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment