mayankshriv commented on pull request #6039: URL: https://github.com/apache/incubator-pinot/pull/6039#issuecomment-700412009
> @mayankshriv in the case of ServiceManager, if I were to add tables as "bootstrap" I would expect those installed before things are healthy, intuitively. This is without regards to later table adds (which would not be bootstrap). Make sense? Yes, I agree with that expectation. Perhaps I don't have a clear picture of the overall solution (since the PR says WIP). So I have a few questions: - I take it that the usage model is only for dev/test environment, and not really being proposed for production. Is this accurate? - Does the proposal involve making changes to the service (to not return healthy status, or something similar), until `ServiceManager` has successfully bootstrapped the tables? If not, great. But if so, this is the part I am having trouble with, as I do not think it is right for a dev/test utility to govern what defines `healthy` status of the service. Also, it opens up questions like, is the service healthy once table is created, or data is uploaded, or data is actually ONLINE in external view to be served (and if so, what happens if one of the segments fails to load). We have similar issue in our integration test, where we have to wait until the table is bootstrapped, before running test queries (eg `OfflineClusterIntegrationTest`). There, the hack is to wait until `count(*)` returns expected count (not necessarily recommending this, just pointing out). ---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org For additional commands, e-mail: commits-h...@pinot.apache.org
