Docker commands take 1 minute to timeout if context endpoint is unreachable #1739
Labels
Comments
|
I suppose the @thaJeztah I will have a look at it |
|
I just looked at it, and it comes from the early call to initializeFromClient in the CLI Initialize function. Actually the problem seems to be present in 18.09 as well: https://github.com/docker/cli/blob/18.09/cli/command/cli.go#L197. I suppose I can work on a fix for the upcoming version, but it might be difficult to backport. |
ijc
pushed a commit
to ijc/docker-cli
that referenced
this issue
May 13, 2019
This partially mitigates docker#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") and is a simpler alternative to docker#1747 (which completely defers the client connection until an actual call is attempted). Note that the previous 60s delay was the culmination of two separate 30s timeouts since the ping is tried twice. This with this patch the overall timeout is 20s. moby/moby#39206 will remove the second ping and once that propagates to this tree the timeout will be 10s. Signed-off-by: Ian Campbell <ijc@docker.com>
thaJeztah
pushed a commit
to thaJeztah/docker
that referenced
this issue
May 13, 2019
When we see an `ECONNREFUSED` (or equivalent) from an attempted `HEAD` on the `/_ping` endpoint there is no point in trying again with `GET` since the server is not responding/available at all. Once vendored into the cli this will partially mitigate docker/cli#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") by cutting the effective timeout in half. Signed-off-by: Ian Campbell <ijc@docker.com> (cherry picked from commit 8c8457b) Signed-off-by: Sebastiaan van Stijn <github@gone.nl>
cpuguy83
pushed a commit
to moby/moby
that referenced
this issue
May 15, 2019
When we see an `ECONNREFUSED` (or equivalent) from an attempted `HEAD` on the `/_ping` endpoint there is no point in trying again with `GET` since the server is not responding/available at all. Once vendored into the cli this will partially mitigate docker/cli#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") by cutting the effective timeout in half. Signed-off-by: Ian Campbell <ijc@docker.com>
docker-jenkins
pushed a commit
to docker-archive/docker-ce
that referenced
this issue
May 15, 2019
When we see an `ECONNREFUSED` (or equivalent) from an attempted `HEAD` on the `/_ping` endpoint there is no point in trying again with `GET` since the server is not responding/available at all. Once vendored into the cli this will partially mitigate docker/cli#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") by cutting the effective timeout in half. Signed-off-by: Ian Campbell <ijc@docker.com> Upstream-commit: 8c8457b0f2f85d58c0e1a5b4a0e174d7808cf886 Component: engine
kiku-jw
pushed a commit
to kiku-jw/moby
that referenced
this issue
May 16, 2019
When we see an `ECONNREFUSED` (or equivalent) from an attempted `HEAD` on the `/_ping` endpoint there is no point in trying again with `GET` since the server is not responding/available at all. Once vendored into the cli this will partially mitigate docker/cli#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") by cutting the effective timeout in half. Signed-off-by: Ian Campbell <ijc@docker.com>
ijc
pushed a commit
to ijc/go-connections
that referenced
this issue
May 17, 2019
This helps to address docker/cli#1739, where an invalid `DOCKER_HOST` setting could result in a 64s delay (that's twice the delay here because the client was trying to hit the `/_ping` endpoint twice, which was addressed in moby/moby#39206) I made a previous attempt to fix this purely on the Docker cli side (docker/cli#1872) however that had the side effect of adding the timeout across the board and not just for the dial phase, which caused a regression for `docker logs -f` (docker/cli#1892) and so was reverted (docker/cli#1893). The new value of 10s is just based on a gut feeling, no initial connection should be taking that long in the real world unless something about the network link is pretty broken (e.g. bad dns perhaps), in which case affected users are surely pretty used to retrying things, better to fail faster in the normal case. Also drop the comment since the linked issue just shows that the original number, just like the new number, was arrived at fairly arbitrarily based on gut feelings (rather than anything empirical) so the reference is not really terribly useful. Signed-off-by: Ian Campbell <ijc@docker.com>
docker-jenkins
pushed a commit
to docker-archive/docker-ce
that referenced
this issue
May 24, 2019
When we see an `ECONNREFUSED` (or equivalent) from an attempted `HEAD` on the `/_ping` endpoint there is no point in trying again with `GET` since the server is not responding/available at all. Once vendored into the cli this will partially mitigate docker/cli#1739 ("Docker commands take 1 minute to timeout if context endpoint is unreachable") by cutting the effective timeout in half. Signed-off-by: Ian Campbell <ijc@docker.com> (cherry picked from commit 8c8457b0f2f85d58c0e1a5b4a0e174d7808cf886) Signed-off-by: Sebastiaan van Stijn <github@gone.nl> Upstream-commit: 61a234d5621cf130db4beb163b8f0b357a27764f Component: engine
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Description
The upcoming context feature seem to cause CLI commands to pause for a 1 minute timeout if the current context's endpoint is not reachable.
Some of my Docker endpoints are only reachable via a VPN connection. If the current context is for an endpoint that I cannot currently reach, every command takes an extra minute (60 seconds) to complete.
Steps to reproduce the issue:
docker context lsordocker info.Describe the results you received:
When the endpoint is unreachable, commands take an extra minute to complete.
In this example, when I switch to
psc-prod-genwithout the requisite network access, Docker CLI commands go through a 1 minute timeout. Even switching back to thedefaultcontext for my local Docker config takes a minute.Describe the results you expected:
Commands should fail immediately if the endpoint is unreachable.
Additional information you deem important (e.g. issue happens only occasionally):
Output of
docker version:Output of
docker info:Additional environment details (AWS, VirtualBox, physical, etc.):
The text was updated successfully, but these errors were encountered: