Apache9 commented on pull request #4202: URL: https://github.com/apache/hbase/pull/4202#issuecomment-1066101926
> Some general thinking about this (not specific to your changes) > > * Could we use the RS start code instead of a new timestamp/seqId? Those should already be unique across RS and that might help in debugging after the fact. > * Do we still need the f1 and f2 files? Could we do away with having two per RS? As explained above, the key here is to make sure that the sequence id of the SFT files written by the new RS is greater than the ones written by the old RS. So the answer of these two questions are pretty easy: 1. No, we can not. The start code of the 'new' RS may be less than the 'old' RS, as here, 'new' and 'old' just mean we reassigned a region from the 'old' to 'new', does not mean the 'new' RS is started after the 'old' RS. 2. To preventing the 'old' RS write SFT files with greater sequence id, we'd better not increase the sequence id when writing them on the same RS, so we'd better still use the old mechanism when writing, i.e, switch between the two SFT files. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
