require operator does not honor past session's success/failure when retrying

I am using require operator to check the status of the previous session of the same workflow. The workflow should -

1. continue if the last session was successful,
2. fail if the last session failed,
3. start last session if the last session does not exist, and,
4. wait for the last session’s completion if it’s pending.

The above four conditions work perfectly when the session is started by the scheduler or is started using digdag start command without the --retry option.

But when I retry a session (either successful or failed) using digdag retry or digdag start --retry or digdag backfill, the require operator starts the last session (condition 3 above) regardless of the status of the last session. In other words, last session is started again even if its status is successful or failed.

[yoyama]

We are developing to enhance require> operator for internal requirements. As described in above comments, retry CLI command generates unique retryAttemptName and it is derived to required WF. As result, the required WF is alway executed. We will change this behavior as stop retryAttemptName derivation. In addition, add option to require> operator to force to run the required WF regardless of attempt exists or not.

[tony810430]

Hi all,

I have encountered this issue recently. Just as @amogh09 said, require operator will check session attempt by both retryAttemptName and sessionTime. That will cause the problem that when I used retry command those related workflows will be executed indeed.

What I expected was retry would only check the latest attempts of require operators and always retry for call operators.

This commit (571cd452c) is a workaround to me. I’m not sure if it is a good solution.

If it looks good to you, I can create a PR for it. What do you think.

Also, cc @frsyuki. Would really appreciate it if you can give me some feedback. Thank you.