[
https://issues.apache.org/jira/browse/TAJO-1466?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14386145#comment-14386145
]
Navis commented on TAJO-1466:
-----------------------------
1. not managed properly
Cache is shared resource by definition but it's not sustaining any sharing
context (reference counter, for example).
2. locks are broken
It's accessed with lock in ExecutionBlockSharedResource, which is created per
execution block. It's not valid because cache is for sharing once-loaded data
between all execution blocks.
3. can cause leakage if cleanup event is missed
Currently, cache is released on state transition but it's critical resource and
should be cleared in usage-block.
> Fix problems in table cache
> ---------------------------
>
> Key: TAJO-1466
> URL: https://issues.apache.org/jira/browse/TAJO-1466
> Project: Tajo
> Issue Type: Bug
> Reporter: Navis
> Assignee: Navis
>
> It's not managed properly and locks on it are broken. And can cause leakage
> if cleanup event is missed.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
