Pull retry policy for pulling remote images

[jccampanero]

The pull request try proposing the definition of the concept of image pull retry policy if an error occurs when pulling remote images.

Try to address #3829 and #6675, allowing both the definition of custom retry policies when dealing with a remote image pull error and provides different retry policies with custom timeouts, number of attempts, etc.

It is a work in progress, test cases should be added: please, consider it a starting point for discussing the proposal.

[linghengqian]

Wasn't this fixed via #9417 ?

[jccampanero]

I apologize for the late reply @linghengqian.

i think that the solution provided by @eddumelendez Edu in #9417 addresses the problem by providing a configurable timeout for retry, and fixing #6675.

But the scope of the PR tries to be more ambitious by defining a configurable policy framework to deal with retries; for instance, it would allow use cases such as the one stated in #3829.

I will try to resolve the conflicts in RemoteDockerImage - sorry, it has been a long time since I proposed the PR and I am afraid it has become obsolete.

Do you think we can work on it in order to include the proposed solution in the project?

Thank you very much for your support.

[linghengqian]

Do you think we can work on it in order to include the proposed solution in the project?

• Unfortunately, I'm not a maintainer of testcontainers; I'm just an external contributor, so I can't offer any reasonable advice.
• It seems most maintainers of testcontainers are Docker employees, and their commits indicate they're currently focused on Docker Desktop, especially the Docker sandbox, and are unlikely to have the time or resources to review the numerous pull requests for testcontainers-java.

[jccampanero]

Do you think we can work on it in order to include the proposed solution in the project?

• Unfortunately, I'm not a maintainer of testcontainers; I'm just an external contributor, so I can't offer any reasonable advice.
• It seems most maintainers of testcontainers are Docker employees, and their commits indicate they're currently focused on Docker Desktop, especially the Docker sandbox, and are unlikely to have the time or resources to review the numerous pull requests for testcontainers-java.

Thank you for letting me know @linghengqian.

In any case, I uploaded a new version of the code, that is consistent with the actual state of the project.

There is only one concern to care about and it is the use of the constant PULL_RETRY_TIME_LIMIT in both DefaultPullRetryPolicy and Awaitility in RemoteDockerImage. I think we can get rid of the constant in the latter, or use a different constant to represent the total time of the pull process, including possible retries.

If there is any interest in the PR, I will try my best to address any possible comments and make the required changes in order to be integrated.

Please, do not hesitate to contact me if we can go on with the PR.

Thank you very much for your support.