Add decompression flag to openstack_images_image_v2 resource

[spnngl]

Some image are compressed and only filling Content-Type header resulting if uploading a compressed image to openstack.

With this we automatically detects and uncompress images in this case.

Example:

$ curl -I "https://stable.release.flatcar-linux.net/amd64-usr/current/flatcar_production_openstack_image.img.bz2"
HTTP/2 200
server: nginx
date: Thu, 16 Feb 2023 15:54:03 GMT
content-type: application/x-bzip2
content-length: 371535602
etag: "rq4xkz657ape"
last-modified: Wed, 15 Feb 2023 18:48:35 GMT
accept-ranges: bytes

Resolves #1481

[kayrus]

thanks for your PR. I left a couple of comments.

[kayrus]

checking for resp.Uncompressed doesn't make sense, since it is used only for Content-Encoding

[spnngl]

I wanted to check if we're not trying to decompress something already decompressed by golang stack or which was never compressed to begin with

[kayrus]

In addition please update the CHANGELOG.md:

## 1.50.0 (Not released yet)

FEATURES

IMPROVEMENTS

* ...

[spnngl]

Did the modifications in the last version

[kayrus]

Thanks. LGTM. Please fix the comments. And I think it's ready to be merged.

[spnngl]

Updated

[kayrus]

@spnngl can you fix golang linter issues?

[kayrus]

It also looks like OpenStack test env doesn't support web-download anymore. Can you add an exception and skip these tests for a while?

[spnngl]

Pushed a new version with lint fix + web_download tests deactivated

[maxime1907]

@kayrus any update? We really need this to be merged! 🙂

[kayrus]

I re-triggered glance tests, once they pass, the PR can be merged.

[kayrus]

can you add a "decompress" exception into the import tests? Tests fail because of

ImportStateVerify attributes not equivalent. Difference is shown below. Top is actual, bottom is expected.

(map[string]string) {
}


(map[string]string) (len=1) {
 (string) (len=10) "decompress": (string) (len=5) "false"
}

terraform-provider-openstack/openstack/import_openstack_images_image_v2_test.go

Lines 29 to 33 in bd2071b

	"region",
	"local_file_path",
	"image_cache_path",
	"image_source_url",
	"verify_checksum",

[spnngl]

I added the decompress ignore + rebase on main

[nikParasyr]

Since changelog is updated this will trigger all the ci jobs.

I think recently opendev,org added some limits to pulling repos etc and as we also have our daily check failing with a timeout. I need to have a look to make sure that the timeout is due to limits and think of a way to bypass it.

Im addressing this in #1497

[kayrus]

@nikParasyr @ozerovandrei I was not aware about the behavior when changelog is changed. I think we can skip failed functional-images test, it has some issues with the data source. other tests are fine.
I think we can merge this PR.

[nikParasyr]

@kayrus the idea behind this behavior with changelog is that we update it only before a release (or after multiple MRs are changed) and the whole ci runs to make sure all services are ok. But it can be a bit annoying if changelog is updated on each MR (also often we have conflicts due to that).

From my side this can be merged as well. The failing ci jobs are not related to this change.