[WIP] Add utility functions for managing containers and images using containers/storage #210

nalind · 2016-11-22T19:52:15Z

To help clean up some of the dependency mess caused by having containers/storage#3 and containers/image#63 calling into each other, this PR takes some of the logic from containers/storage#3 and moves it here. If we merge it, we'll want to adapt #189 to call those routines from here.

nalind · 2016-11-22T19:55:00Z

@mrunalp, @runcom, this should keep containers/storage from needing to grow a dependency on containers/image.

runcom · 2016-11-22T20:22:17Z

Could we move this under ./pkg/storage? Thet pkg directory will be the house for future refactor to abstract useful code (instead of cluttering the main directory layout). Wdyt?

nalind · 2016-11-22T20:36:01Z

Okay, moved it.

runcom · 2016-11-26T12:59:09Z

Make lint and tests are failing, also we should probably remove linting vendors here.

mrunalp · 2016-11-28T20:34:07Z

I have created #216 for skipping vendor but there are still lint errors.

Crazykev · 2016-11-29T07:01:59Z

pkg/storage/utils.go

+		return true
+	}
+	for _, name := range names {
+		if filter == "" || filter == name {


Why add a condition filter == "" here? I mean you have checked filter is not empty string in ListImage, and if we really need to check it here, why not together with filter == ID, IIRC filter == "" means no filter at all, maybe don't need to check it for every name.

Good point. Updating.

Crazykev · 2016-11-29T07:05:30Z

pkg/storage/utils.go

+			return true
+		}
+	}
+	if len(filter) < 7 {


Is there any constrain for "7"?

It's a placeholder value, chosen mainly because that's the length that container and image IDs are often truncated for display by dockerd. A longer version would probably use a truncindex, since all we're really looking for is a sufficiently-long value to be unambiguous about which image or container or pod the caller is referring to. I'll see about switching to doing that.

👍 to using truncIndex.

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind · 2016-12-06T20:59:18Z

Okay, updated the ListImages() logic to assume that the store's GetImage() method will begin to accept truncated IDs, which is where adding a truncIndex to it will lead (essentially, it'll treat truncated IDs as a different sort of name, resolving the passed-in value to a full-length image ID before proceeding). A non-empty filter will now select the image with an exactly-matching name or ID, or truncated ID once that feature lands there, and an empty filter will continue to return a list of all images.

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

mrunalp · 2016-12-09T00:39:26Z

make fails for me locally:

mkdir -p /root/gosrc/src/github.com/kubernetes-incubator/cri-o/vendor/src/github.com/kubernetes-incubator
ln -sfn /root/gosrc/src/github.com/kubernetes-incubator/cri-o /root/gosrc/src/github.com/kubernetes-incubator/cri-o/vendor/src/github.com/kubernetes-incubator/cri-o
go build --tags "selinux" -o ocid ./cmd/server/
vendor/src/github.com/containers/image/docker/docker_image_src.go:16:2: cannot find package "github.com/docker/distribution/registry/client" in any of:
        /usr/lib/golang/src/github.com/docker/distribution/registry/client (from $GOROOT)
        /root/gosrc/src/github.com/kubernetes-incubator/cri-o/vendor/src/github.com/docker/distribution/registry/client (from $GOPATH)
make: *** [Makefile:49: ocid] Error 1

mrunalp · 2016-12-09T00:42:24Z

pkg/storage/image.go

+	GetStore() storage.Store
+}
+
+func (svc *imageService) ListImages(ctx context.Context, filter string) ([]ImageResult, error) {


ctx isn't used in this function.

mrunalp · 2016-12-09T00:43:00Z

pkg/storage/image.go

+	return results, nil
+}
+
+func (svc *imageService) ImageStatus(ctx context.Context, nameOrID string) (*ImageResult, error) {


ctx isn't used in this function.

mrunalp · 2016-12-09T00:43:30Z

pkg/storage/image.go

+	return nil
+}
+
+func (svc *imageService) PullImageUsingContexts(ctx context.Context, imageName string, policyContext *signature.PolicyContext, options *copy.Options) (types.ImageReference, error) {


ctx isn't used in this function.

mrunalp · 2016-12-09T00:44:01Z

pkg/storage/image.go

+		return nil, err
+	}
+	options := copy.Options{}
+	ref, err := svc.PullImageUsingContexts(ctx, imageName, policyContext, &options)


ctx passed down but not used in the function being called.

mrunalp · 2016-12-09T00:44:25Z

pkg/storage/image.go

+	return ref, nil
+}
+
+func (svc *imageService) RemoveImage(ctx context.Context, nameOrID string) error {


ctx isn't used in this function.

Right, that's a leftover from when I thought we'd need to pass it to the image library. Removing it.

mrunalp · 2016-12-09T00:46:53Z

pkg/storage/runtime.go

+	layerName := metadata.ContainerName + "-layer"
+	names, err = r.image.GetStore().GetNames(container.LayerID)
+	if err != nil {
+		cleanup(container)


Should the cleanup function be deferred instead so it is called on all subsequent error exits?

Might as well. Changing.

runcom · 2016-12-09T13:37:51Z

pkg/storage/image.go

+	PullImage(ctx context.Context, imageName string) (types.ImageReference, error)
+	// RemoveImage deletes the specified image.
+	RemoveImage(ctx context.Context, imageName string) error
+	GetStore() storage.Store


missing comment on GetStore()

runcom · 2016-12-09T13:42:26Z

@mrunalp I'm not getting any errors 😕

14:41:25 [github.com/kubernetes-incubator/cri-o] ‹pr/origin/210› make binaries
go build --tags "selinux seccomp  " -o ocid ./cmd/server/
go build -o ocic ./cmd/client/
go build -o kpod ./cmd/kpod/
make -C conmon
make[1]: Entering directory '/home/amurdaca/go/src/github.com/kubernetes-incubator/cri-o/conmon'
make[1]: 'conmon' is up to date.
make[1]: Leaving directory '/home/amurdaca/go/src/github.com/kubernetes-incubator/cri-o/conmon'
make -C pause
make[1]: Entering directory '/home/amurdaca/go/src/github.com/kubernetes-incubator/cri-o/pause'
make[1]: 'pause' is up to date.
make[1]: Leaving directory '/home/amurdaca/go/src/github.com/kubernetes-incubator/cri-o/pause'

runcom · 2016-12-09T13:43:33Z

BTW, this looks good to me overall (when mrunal's comments are addressed also) - I'm testing the whole storage PRs with k8s and so far they looks great. We can always reiterate when we have those merged I guess.

Based on review comments from cri-o/cri-o#210: * use a deferred anonymous function to do cleanup if creating a container fails, instead of explicitly calling a local function * add missing godoc string for ImageServer.GetStore * remove context.Context arguments for ImageServer and RuntimeServer methods, which we ended up never using anyway Signed-off-by: Nalin Dahyabhai <[email protected]>

Based on review comments from cri-o#210: * use a deferred anonymous function to do cleanup if creating a container fails, instead of explicitly calling a local function * add missing godoc string for ImageServer.GetStore * remove context.Context arguments for ImageServer and RuntimeServer methods, which we ended up never using anyway Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind · 2016-12-15T18:29:23Z

Updated to pull in the current versions of containers/storage and containers/image.

runcom · 2016-12-15T18:37:02Z

I'll test this out soon /cc @mrunalp

runcom · 2016-12-15T23:23:34Z

@kubernetes-incubator/maintainers-cri-o PTAL as well

mrunalp · 2016-12-16T22:09:00Z

@nalind Could you clean up the commit history?

nalind · 2016-12-16T22:19:51Z

@mrunalp Sure. Do you mean by squashing to a single patch, or expanding in commit logs, or something else?

mrunalp · 2016-12-16T22:23:31Z

@nalind I don't mind multiple commits but with a clean history such that we don't change code added in an earlier commit to fix something within the same PR. Also, we are trying to keep the history bisectable so tests could pass on each commit.

nalind · 2016-12-16T22:24:10Z

Okay, that's going to take some time, but I'll work on it.

mrunalp · 2016-12-16T22:28:16Z

@nalind okay, thanks!

nalind · 2016-12-18T04:45:28Z

@mrunalp That basically entailed squashing most of the code changes into one patch, and reordering a couple of things. Hopefully it's not too spare.

mrunalp · 2016-12-18T05:49:05Z

No worries. Thanks!

…

On Dec 17, 2016, at 8:45 PM, Nalin Dahyabhai ***@***.***> wrote: @mrunalp That basically entailed squashing most of the code changes into one patch, and reordering a couple of things. Hopefully it's not too spare. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub, or mute the thread.

runcom · 2016-12-19T16:42:49Z

@nalind could you rebase this? I believe we're good to go with this as the building blocks for #189
we'll re-iterate if anything wasn't super perfect but it's clear storage is critical to CRI-O in order to remove the docker dependency we have right now.

@mrunalp PTAL also

Add the necessary build tags and configuration so that integration tests can properly build against device mapper and btrfs libraries. Signed-off-by: Nalin Dahyabhai <[email protected]>

Update the versions of containers/storage and containers/image, and add new dependencies that they pull in. Signed-off-by: Nalin Dahyabhai <[email protected]>

Any binary that will be managing storage needs to initialize the reexec package in order to be able to apply or read image layers. Signed-off-by: Nalin Dahyabhai <[email protected]>

Add an intermediate API layer that uses containers/storage, and a containers/image that has been patched to use it, to manage images and containers, storing the data that we need to know about containers and pods in the metadata fields provided by containers/storage. While ocid manages pods and containers as different types of items, with disjoint sets of IDs and names, it remains true that every pod includes at least one container. When a container's only purpose is to serve as a home for namespaces that are shared with the other containers in the pod, it is referred to as the pod's infrastructure container. At the storage level, a pod is stored as its set of containers. We keep track of both pod IDs and container IDs in the metadata field of Container objects that the storage library manages for us. Containers which bear the same pod ID are members of the pod which has that ID. Other information about the pod, which ocid needs to remember in order to answer requests for information about the pod, is also kept in the metadata field of its member containers. The container's runtime configuration should be stored in the container's ContainerDirectory, and used as a template. Each time the container is about to be started, its layer should be mounted, that configuration template should be read, the template's rootfs location should be replaced with the mountpoint for the container's layer, and the result should be saved to the container's ContainerRunDirectory, for use as the configuration for the container. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind · 2016-12-19T17:03:08Z

Okay, rebased.

mrunalp · 2016-12-19T18:10:39Z

LGTM

runcom · 2016-12-19T18:12:48Z

let's get this in 🎉

rhatdan · 2016-12-19T18:24:45Z

Great work Nalin

We've been hardcoding it since 0e4af6d (*: add seccomp buildtag, 2016-11-29, cri-o#219), but that predates the hack/* approach which landed in f893e38 (Add build tags for integration tests, 2016-12-07, cri-o#210). I'm not sure why cri-o#210 skipped libseccomp (possibly because we always install it on Linux in .travis.yml), but now that we are trying to build on OS X too we need to be more flexible.

We've been hardcoding it since 0e4af6d (*: add seccomp buildtag, 2016-11-29, cri-o#219), but that predates the hack/* approach which landed in f893e38 (Add build tags for integration tests, 2016-12-07, cri-o#210). I'm not sure why cri-o#210 skipped libseccomp (possibly because we always install it on Linux in .travis.yml), but now that we are trying to build on OS X too we need to be more flexible. Signed-off-by: W. Trevor King <[email protected]>

Update variables needed by osbuilder. Also fix query to get the initrd base OS. Fixes: cri-o#210 Signed-off-by: Jose Carlos Venegas Munoz <[email protected]>

k8s-ci-robot added the cncf-cla: yes label Nov 22, 2016

nalind mentioned this pull request Nov 22, 2016

Do not merge: additional support routines for CRI use cases containers/storage#3

Closed

nalind force-pushed the metadata branch from 9fe5eb2 to 87f3aaf Compare November 22, 2016 20:35

Crazykev reviewed Nov 29, 2016

View reviewed changes

nalind force-pushed the metadata branch from 87f3aaf to 6003a4f Compare December 5, 2016 19:49

nalind added a commit to nalind/storage that referenced this pull request Dec 6, 2016

Avoid a spurious empty string check

8979419

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind added a commit to nalind/storage that referenced this pull request Dec 6, 2016

Avoid a spurious empty string check

4913823

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind force-pushed the metadata branch from 6003a4f to 7155534 Compare December 6, 2016 20:58

nalind added a commit to nalind/storage that referenced this pull request Dec 6, 2016

Avoid a spurious empty string check

b40b7ac

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind added a commit to nalind/storage that referenced this pull request Dec 8, 2016

Avoid a spurious empty string check

63c7244

Spotted in review of cri-o/cri-o#210 by Crazykev. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind force-pushed the metadata branch from c522e92 to 772e4f4 Compare December 8, 2016 17:52

mrunalp reviewed Dec 9, 2016

View reviewed changes

runcom reviewed Dec 9, 2016

View reviewed changes

nalind force-pushed the metadata branch from 772e4f4 to 53da9dd Compare December 9, 2016 18:30

nalind force-pushed the metadata branch from aaff1f3 to 48abcfa Compare December 15, 2016 18:08

runcom mentioned this pull request Dec 16, 2016

Storage in Image Management #189

Merged

runcom added this to the 0.0.1 milestone Dec 17, 2016

nalind force-pushed the metadata branch from 48abcfa to 95f988d Compare December 18, 2016 04:37

runcom mentioned this pull request Dec 19, 2016

Add support kpod list images #225

Closed

nalind added 4 commits December 19, 2016 11:44

Add build tags for integration tests

f893e38

Add the necessary build tags and configuration so that integration tests can properly build against device mapper and btrfs libraries. Signed-off-by: Nalin Dahyabhai <[email protected]>

Update containers/storage and containers/image

9b88295

Update the versions of containers/storage and containers/image, and add new dependencies that they pull in. Signed-off-by: Nalin Dahyabhai <[email protected]>

Initialize the reexec package

d45ff58

Any binary that will be managing storage needs to initialize the reexec package in order to be able to apply or read image layers. Signed-off-by: Nalin Dahyabhai <[email protected]>

nalind force-pushed the metadata branch from 95f988d to 4ae8606 Compare December 19, 2016 16:55

runcom merged commit a719085 into cri-o:master Dec 19, 2016

egernst pushed a commit to egernst/cri-o that referenced this pull request Nov 26, 2018

image: Fix image build

47010dd

Update variables needed by osbuilder. Also fix query to get the initrd base OS. Fixes: cri-o#210 Signed-off-by: Jose Carlos Venegas Munoz <[email protected]>

[WIP] Add utility functions for managing containers and images using containers/storage #210

[WIP] Add utility functions for managing containers and images using containers/storage #210

Uh oh!

Conversation

nalind commented Nov 22, 2016

Uh oh!

nalind commented Nov 22, 2016

Uh oh!

runcom commented Nov 22, 2016

Uh oh!

nalind commented Nov 22, 2016

Uh oh!

runcom commented Nov 26, 2016

Uh oh!

mrunalp commented Nov 28, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

nalind commented Dec 6, 2016

Uh oh!

mrunalp commented Dec 9, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

runcom commented Dec 9, 2016

Uh oh!

runcom commented Dec 9, 2016

Uh oh!

nalind commented Dec 15, 2016

Uh oh!

runcom commented Dec 15, 2016

Uh oh!

runcom commented Dec 15, 2016

Uh oh!

mrunalp commented Dec 16, 2016

Uh oh!

nalind commented Dec 16, 2016

Uh oh!

mrunalp commented Dec 16, 2016

Uh oh!

nalind commented Dec 16, 2016

Uh oh!

mrunalp commented Dec 16, 2016

Uh oh!

nalind commented Dec 18, 2016

Uh oh!

mrunalp commented Dec 18, 2016 via email

Uh oh!

runcom commented Dec 19, 2016

Uh oh!

nalind commented Dec 19, 2016

Uh oh!

mrunalp commented Dec 19, 2016

Uh oh!