[
https://issues.jboss.org/browse/TEIID-2475?page=com.atlassian.jira.plugin...
]
Steven Hawkins resolved TEIID-2475.
-----------------------------------
Fix Version/s: 8.4.1
8.5
(was: 8.4)
(was: 7.7.8)
Resolution: Done
Changed from reporting "node process time" to "node next batch process
time".
The "cumulative node process time" overlaps for child nodes - since it starts
when a batch is first fetched and the blocked exception may be caught. So it is generally
not possible to compute just the "node process time", we do know however that
the child next batch times do not overlap, so we can somewhat accurately report just the
"node next batch process time".
Another issue seen but not addressed is that nodes that directly return a tuple buffer
(such as a full sort) will not report any node stats as getNextBatch is never consulted.
I'm not sure what should be done in these case as we don't wont to needlessly
iterate over the batches.
The RelationalNode.collectNodeStats is only subtracting out the last
node
-------------------------------------------------------------------------
Key: TEIID-2475
URL:
https://issues.jboss.org/browse/TEIID-2475
Project: Teiid
Issue Type: Bug
Components: Query Engine
Affects Versions: 6.0.0
Reporter: Debbie Steigner
Assignee: Steven Hawkins
Fix For: 8.4.1, 8.5
Attachments: Query_Plan.txt
The "Node Cumulative Next Batch Process Time" Statistic
If you have a look at the explain plan (attached). Following the definition,
the JoinNode Node Cumulative Next Batch Process Time: 565471. So, it is the summation of
AccessNode (Node Process Time: 892860) + DependentAccessNode (Node Process Time: 439) +
JoinNode (Node Process Time: 2203120). The summation does not agree.
The RelationalNode.collectNodeStats looks wrong (it's effectively only subtracting
out the last node)
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see:
http://www.atlassian.com/software/jira