Github user keith-turner commented on the issue:
https://github.com/apache/accumulo/pull/134
I have been looking at the PR and I have come to two conclusions. First,
I don't think these change will negatively impact the normal single threaded
case in any way. Second, I can't come the conclusion that these changes are
completely correct for the multi-threaded case because I don't have a good
sense of what specific problems are happening in the multi-threaded case. In
order to reason if this PR solves the problems, I need to know what the
problems are.
I do think these changes are a good improvement and are ready to merge.
They definitely solve some subset of multi-threaded problems, just not sure if
all.
One of thing I thought of that may be useful to try in the test is having
threads close RFiles while other threads are reading from deep copies. Doing
this in a continuous manner with multiple RFile concurrently, such that some
threads are opening and reading rfiles and other threads are closing those
rfiles. This would attempt to force multiple threads to use the same
decompressor.
---
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.
---
