Modify docker volume inspect to return existed volumes and the names of the unexsited volumes
Upstream-commit: 5b4734aaa543243ba076bf606286bbc46fed9e38
Component: engine
It's in .gitignore too, so why not
Signed-off-by: Alexander Morozov <lk4d4@docker.com>
Upstream-commit: 9c8d6edbf15a33045cffb37b6406715dc8776675
Component: engine
Removed images were not cleaned up from the
digest-set that is used for the search index.
Fixes#18437
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
Upstream-commit: fcb083c6ac0fdb71e8db7a58f07a561342d631d3
Component: engine
We were calling Stat for each layer to get the size so we could indicate
progress, but https://github.com/docker/distribution/pull/1226 made it
possible to get the length from the GET request that Open initiates.
Saving one round-trip per layer should make pull operations slightly
faster and more robust.
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Upstream-commit: 39589800b4750bf28078efe57f7f1e74d971248f
Component: engine
This solves a bug where /etc may have pre-existing permissions from
build time, but init layer setup (reworked for user namespaces) was
assuming root ownership. Adds a test as well to catch this situation in
the future.
Minor fix to wrong ordering of chown/close on files created during the
same initlayer setup.
Docker-DCO-1.1-Signed-off-by: Phil Estes <estesp@linux.vnet.ibm.com> (github: estesp)
Upstream-commit: 23b771782ab7236ce5024ac5773a6ded9a2af753
Component: engine
This function was starting a goroutine that modifies one of its return
values. The intent is for the goroutine to only influence the return
value when it's causing the function to return, but it's racy and can
also modify the return value when the function is returning due to the
timeout. Fix the goroutine to not modify return values directly.
Also, give the channel a buffer so that the goroutine doesn't block
forever after a timeout.
Fixes#18305
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Upstream-commit: 2704fd9156bfb0fb8dc16c42902bb18ea5aa94a9
Component: engine
I saw a failure of TestDockerCmdWithTimeout. This test starts a command
that produces output after 10 ms, but uses a 5 ms timeout, so normally
the command will be killed before the output. The time intervals are so
small that the timeout may not reliably trigger before the output, which
can cause the test to fail.
This commit changes the test to only fail if the process is still alive
after 10 seconds. This means the test will confirm that the timeouts are
happening, but not attempt to gauge that the timeouts are happening
within milliseconds of when they are expected (which can't be done
reliably).
Signed-off-by: Aaron Lehmann <aaron.lehmann@docker.com>
Upstream-commit: 13d768b8ee05d3158c62d761e4ebe657cd7a8c25
Component: engine
With content addressability update starting upgraded
daemon for the first time can take a long time if
graph dir was not prepared with a migration tool before.
This avoids systemd timeouts while the migration is
taking place.
Signed-off-by: Tonis Tiigi <tonistiigi@gmail.com>
Upstream-commit: eff5e6498fa424587c339b5ae54e28f003775e4b
Component: engine
This change allows API clients to retrieve an authentication token from
a registry, and then pass that token directly to the API.
Example usage:
REPO_USER=dhiltgen
read -s PASSWORD
REPO=privateorg/repo
AUTH_URL=https://auth.docker.io/token
TOKEN=$(curl -s -u "${REPO_USER}:${PASSWORD}" "${AUTH_URL}?scope=repository:${REPO}:pull&service=registry.docker.io" |
jq -r ".token")
HEADER=$(echo "{\"registrytoken\":\"${TOKEN}\"}"|base64 -w 0 )
curl -s -D - -H "X-Registry-Auth: ${HEADER}" -X POST "http://localhost:2376/images/create?fromImage=${REPO}"
Signed-off-by: Daniel Hiltgen <daniel.hiltgen@docker.com>
Upstream-commit: 8dce8e9901501863a80caf47d71713d4d36925a3
Component: engine
