keith-turner commented on issue #3823:
URL: https://github.com/apache/accumulo/issues/3823#issuecomment-1782952551
For concurrent merge and bulk import, thought of the following strategy that
I like.
* Merge does not acquire opid when loaded flags are present, using
conditional mutations.
* When the merge code is in the process of acquiring opids and observes
loaded flags, it relinquishes the opids it has obtained. This is to allow for
overlapping bulk imports to load into those tablets.
With the above strategy bulk import and merge will not run concurrently.
Also bulk import does not have to take any extra step, only the merge code.
So since bulk import will run much more frequently than merge, its good to not
add extra steps to bulk import to handle a rare condition. One downside with
this strategy is that its possible that merge could never run if there are
continuous bulk imports.
--
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]
