Getting failed build due to "Github or AWS" circle error every 1 out of 2 builds

This is a very consistent issue. Exactly every other build, I get that it was canceled right after trying to restore cache, which there is nothing to restore. And when it fails, I get this error: “Looks like we had a bug in our infrastructure, or that of our providers (generally GitHub or AWS) We should have automatically retried this build. We’ve been alerted of the issue and are almost certainly looking into it, please contact us if you’re interested in the cause of the problem.” I messaged this to support at first and received a cut and paste response basically saying that it was my problem. Thankfully after that I got someone else to reply who told me to open a ticket here. Please let me know if there is any other info I could provide.

Are you caching any directories outside the home folder? We cannot actually restore anything that is located outside of the home folder.

If that’s the case, please place the files you would like to cache into the home folder and then symlink them into the right place during the build, after the cache has been restored (dependencies step, for example).

I said that there’s nothing to restore. I’m not caching anything.

And even if I was caching, I’d be very surprised that it works exactly every other time.

Sorry for the delay here @cpg1111. Is this still happening for you? Would it be an option for you to paste the whole circle.yml file right here? If you are alright with sharing a build URL (if this is a private project, you might not want to do that), please post it here as well. Thanks.

Sorry, this is a private project, but what I can tell you is that is breaks right after the machine starts up docker, as in the daemon.

Alright, thanks for that. Is this an OS X / iOS build? If so, Docker is still not supported on the OS X machines.

Also, what is the next step in your build that comes after the Docker start-up? It might be that part that’s causing the issue.

…um no it’s a Linux build, otherwise I wouldn’t be using docker. The next step is pulling the last successful build of the containers from my docker registry, then building the new ones.

Thanks for the details. I’ll have to look at the logs on our side to figure out what is going on here. Could you please contact us via in-app support mentioning this post and a link to the build that shows the failure? I’ll check the logs and we’ll continue the conversation here on Discuss. Thanks.

1 Like