chenjunjiedada commented on issue #374: Migrate spark table to iceberg table URL: https://github.com/apache/incubator-iceberg/pull/374#issuecomment-522904779 @rdblue, @aokolnychyi @jerryshao PR updated. It uses `coalesce(1)` to append all files to one manifest file. In the case of a large table which has a mass of files, the final manifest may be too big. Currently, I'm not sure what is the best way to transfer manifests between the executor and the driver. IIUC, there are two ways if we want to build manifests parallel and commit once. 1. Store the temporary manifests to some HDFS location and return location string to the driver, then the driver constructs the manifests and appends manifests and commits. While I don't find an existing API to build manifests from location, ManifestReader? 2. Implement `Serializable ` for the Manifest or Snapshot, so that we collect them and append manifests in the driver side. While this looks like a bit complex. What do you think?
---------------------------------------------------------------- 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: [email protected] With regards, Apache Git Services --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
