fatal error: all goroutines are asleep - deadlock!

[bep]

fatal error: all goroutines are asleep - deadlock!

[bep]

I have pushed a fix that should fix the deadlock issue. Note that I suspect this situation only happens in error situations, so your problem is most likely not 100% fixed, but now it should be easier to debug.

[bep]

And please let me know if this helped -- if you don't see any error in the console, I may have to dig some more.

[01AutoMonkey]

Using the latest version I get:

s3deploy 1.1.1, commit 06dd326d8838bbf4f62304af49f3a1803a99bed1, built at 2017-11-30T19:12:40Z
Deleted: 0 Uploaded: 0 Skipped: 0 (NaN% changed)

I just cd into my hugo site directory which contains a public/ and execute the command like so (excluding bucket name, key, and secret from example): s3deploy -v -source=public/ -region=eu-west-2 -bucket=<mybucketname> -key=<mykey> -secret=<mysecret>.

I've also set the IAM action permissions for my bucket, just a copy/paste from the README.md and inserted my actual bucket name.

I've tried it with and without a .s3deploy.yml file.

Forestry.io has been handling my deploys up until now, so it's not an empty bucket, but I've also tried deleting all objects in my bucket before executing the command.

[bep]

This is hard for me to debug with what you say. I have not experienced this, and I have used it a lot.

May be a path issue. Are you on Windows?

[bep]

@01AutoMonkey see #8 -- while it may not solve your issue, I notice that there are some errors that should be handled better -- I will fix that.

[01AutoMonkey]

I'm on Ubuntu and I've tried all variations of the source path (absolute path, ~/... path, relative path).

Also, playing around with the code I notice d.FileCount() returns 0, which may explain the (NaN% changed).

[bep]

That is weird, I have several sites that deploys from CircleCI on LInux, e.g.:

https://github.com/bep/bego.io/blob/master/circle.yml

I will do more on the error handling on this in a day or two.

[01AutoMonkey]

Adding print like so:

	// List all files in the remote bucket
	contents, err := destBucket.GetBucketContents()
	if err != nil {
		fmt.Println(err)
		return stats, err
	}

Results in:

s3deploy dev, commit none, built at unknown
Get /test.bucket.is/?delimiter=&marker=&max-keys=1000&prefix=: unsupported protocol scheme ""
Deleted: 0 Uploaded: 0 Skipped: 0 (0% changed)

Which lead me to realize that eu-west-2 isn't part of aws.Regions (maybe it's too new?), if I replace it with eu-west-1 (which is part of aws.Regions) I get a slightly different error:

s3deploy dev, commit none, built at unknown
Get https:/test.bucket.is/?delimiter=&marker=&max-keys=1000&prefix=: 301 response missing Location header
Deleted: 0 Uploaded: 0 Skipped: 0 (0% changed)

The URL is now a bit better but not quite (though it would be wrong anyway given my actual region is eu-west-2.

This is most likely the same error Alan_F got at: https://discourse.gohugo.io/t/howto-deploying-hugo-on-s3-and-cloudfront/2800/26

[bep]

The solution is probably in #3