Skip to content

[bug] container not following SIGTERM signals to close down (inc possible fixes) #21

Closed
@andrewcrook

Description

@andrewcrook

Hello,

I keep getting the following error Docker container borg-server stopped unexpectedly around the time watchtower is scheduled. I have looked into and the only thing I think that has happened is that the container isnt responding to SIGTERM signals causing watchtower, after sometime, to force it via SIGKILL.

I read that this can be caused if the processes of the container are called under bash -c because the SIGTERM signals are not passed on to processes.

Apparently
ENTRYPOINT /run.sh
wraps /run.sh in bash -c

and

ENTRYPOINT [“/run.sh”]
doesn’t
as officially explained

You can also propagate SIGTERM signals to bash child processes using bash’s trap and wait

Lastly, not all docker containers or applications use SIGTERM here you can use dockerfile's STOPSIGNAL to change the signal type

I think its probably the first two issues.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions