The Wayback Machine - https://web.archive.org/web/20201104144518/https://github.com/clearcontainers/runtime/issues/467
Skip to content
This repository has been archived by the owner. It is now read-only.

docker stop showing 'errors' in runtime log #467

Open
grahamwhaley opened this issue Aug 25, 2017 · 6 comments
Open

docker stop showing 'errors' in runtime log #467

grahamwhaley opened this issue Aug 25, 2017 · 6 comments

Comments

@grahamwhaley
Copy link
Contributor

@grahamwhaley grahamwhaley commented Aug 25, 2017

Distro: Ubuntu 16.04.2 LTS server
cc-runtime: 3.0.0-beta1 (from OBS install)

Running:

docker run -tid busybox sh
docker stop XXXXXX
docker rm XXXXXXX

where XXXXXX is the container ID from the first step, is producing two errors for me in the runtime log (that I enabled by uncommenting the relevant section of the /etc... config file):

2017-08-25 15:41:29.834518887 +0100 BST:2952:cc-runtime:error:Container cc80658aae0bad16ad5c1be722b0d06dc54c8b5d6978455a0f14a684bc9b65bd not ready or running, cannot send a signal
2017-08-25 15:41:32.750079653 +0100 BST:2968:cc-runtime:error:Container ID does not exist

The ID there matches the XXXXXX above.

afaict, the error lines appear during the docker rm phase - but, I am not wholly convinced that is right.
I would post the whole/rest of the relevant bits of the log, but it is about 800 long lines, so perhaps I'll not do that yet.

@egernst
Copy link
Member

@egernst egernst commented Aug 30, 2017

@grahamwhaley -- I see the same thing when doing:
docker run -it alpine sh

exit

This is a bit concerning, especially given some of the other life cycle bugs we have open.

@jodh-intel
Copy link
Contributor

@jodh-intel jodh-intel commented Aug 31, 2017

Closing as this isn't a bug in the runtime, it's a bug in containerd afaik:

And here's the relevant snippets from a debug log showing how the runtime was called for a single container lifecycle:

$ sudo grep "called as" /var/lib/clear-containers/runtime/runtime.log|cut -d\) -f2-|tr -d :|tr -d \[|awk '{print $1,$2,$3}'
called as create
called as start
called as kill
called as delete
called as delete
@jodh-intel jodh-intel closed this Aug 31, 2017
@grahamwhaley
Copy link
Contributor Author

@grahamwhaley grahamwhaley commented Aug 31, 2017

So @jodh-intel , should we be changing our error to a warning, to match cc-oci-runtime better (and cause less concerning messages in the logs) ?
If so, we might want to re-open this or a new Issue.

@jodh-intel
Copy link
Contributor

@jodh-intel jodh-intel commented Aug 31, 2017

It's very debatable. The runtime has detected that it has been called incorrectly, so that is an error. I'm easy either way though as long as we log at atleast the warning level.

@grahamwhaley
Copy link
Contributor Author

@grahamwhaley grahamwhaley commented Aug 31, 2017

OK, after irc chatter and input from @egernst I'm going to re-open this asking that we downgrade those errors to info/warnings - given that the double delete is, I believe, benign, correct @jodh-intel ?

@grahamwhaley grahamwhaley reopened this Aug 31, 2017
@jodh-intel
Copy link
Contributor

@jodh-intel jodh-intel commented Aug 31, 2017

It's benign in that the runtime catches the incorrect call from containerd and handles it, yes.
@egernst - please can you reference this bug when you raise an issue on containerd.

@jodh-intel jodh-intel added the bitesize label Aug 31, 2017
mcastelino pushed a commit to mcastelino/runtime that referenced this issue Dec 6, 2018
Once `containerID` and `sandboxID` fields are available, re-register
the logger with the external packages to ensure they too display these
important fields.

Fixes clearcontainers#467.

Signed-off-by: James O. D. Hunt <james.o.hunt@intel.com>
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
3 participants
You can’t perform that action at this time.