Github user denine99 commented on the pull request:
https://github.com/apache/accumulo/pull/22#issuecomment-84524662
I fixed up the patch from Keith's comments. This one looks good.
My mistake starting out with the patch was trying to change *every*
location where an empty tablet could be disregarded during compaction. Now the
current patch has the *minimal* changes necessary to make empty tablets
compact. The biggest one is the change to the default compaction strategy. As
before, I'm open for ideas to test further. I think we're closing in on good
code.
To answer the question I posed previously, the expected behavior when a
major compaction iterator throws and exception (e.g.
[BadIterator](https://accumulo.apache.org/1.6/apidocs/org/apache/accumulo/test/functional/BadIterator.html))
is that the tablet server will fail the compaction, not updating the
metadata table, and then retry the compaction. If the error is permanent, then
the compaction will fail again and again continuously. The only way to stop
the cycle is to cancel the compaction manually. This patch maintains this
behavior.
Rebased the commits to master. Will squash once given a green light.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
