[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13913302#comment-13913302
]
Mike Matrigali commented on DERBY-2130:
---
consider for 10.10 backport
Optimizer
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13913337#comment-13913337
]
Mike Matrigali commented on DERBY-2130:
---
looking at backporting this to 10.10
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13913749#comment-13913749
]
ASF subversion and git services commented on DERBY-2130:
Commit
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13803544#comment-13803544
]
Kathey Marsden commented on DERBY-2130:
---
I tried patched 10.2 and found the prepare
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13800791#comment-13800791
]
Kathey Marsden commented on DERBY-2130:
---
So should I go ahead and check this in as
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13801328#comment-13801328
]
Bryan Pendleton commented on DERBY-2130:
+1 to checking in the patch. It clearly
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13801431#comment-13801431
]
ASF subversion and git services commented on DERBY-2130:
Commit
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13800149#comment-13800149
]
Kathey Marsden commented on DERBY-2130:
---
Verified this is still a problem. On my
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13800159#comment-13800159
]
Bryan Pendleton commented on DERBY-2130:
Three times faster is nothing to sneeze
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13464757#comment-13464757
]
Bryan Pendleton commented on DERBY-2130:
I don't consider the patch high risk. I
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=13464142#comment-13464142
]
Kathey Marsden commented on DERBY-2130:
---
Is the patch considered high risk? Would
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12859413#action_12859413
]
Rick Hillegas commented on DERBY-2130:
--
Hi Bryan,
You contributed a patch for this
[
https://issues.apache.org/jira/browse/DERBY-2130?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanelfocusedCommentId=12859624#action_12859624
]
Bryan Pendleton commented on DERBY-2130:
Hi Rick,
I think that Army's patch is a
Bryan Pendleton commented on DERBY-2130:
With jumpReset.patch applied, I cannot reproduce the varying optimize times.
The times are all in a tight range, after nearly 6x as many tests as produced
the variable optimize times before.
That's good to hear.
I'm curious: what is the optimization time that you consistently see
with jumpReset.patch? Is it the lesser time (i.e 170-200 seconds) or
the greater time (500-650s)? Or something else entirely?
The values I saw on the machine in question were consistently 192-198 secs.
That said, I would
Bryan Pendleton wrote:
Independently, I tried fiddling with the cost estimates myself, because as
I stepped through HeapCostController and BTreeCostController, my reaction
was that the costs were possibly 2 orders of magnitude too high. In my
experiments, I didn't see much benefit to the
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12456541 ]
A B commented on DERBY-2130:
Is it possible that the 10.1 vs 10.3 database is a red herring here,
and this is simply the same non-determinism that I was observing
A B commented on DERBY-2130:
Have you yourself seen this variance with *10.1* databases, as well?
No, but it hadn't occurred to me to even try the experiment until
you mentioned it :)
I'll try to confirm or disprove this theory, but I'm not able to
spend as much
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12456115 ]
Bryan Pendleton commented on DERBY-2130:
When running against a 10.1 database 10.3 estimates a cost of 1736685,
followed up by a
cost of 198589 for
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12455802 ]
A B commented on DERBY-2130:
Thanks for the query plans, Bryan. I took a quick look and nothing stood out,
but I'll have to look more closely before I can say
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12455197 ]
Bryan Pendleton commented on DERBY-2130:
So far as I know, the jumping code is working properly.
Thanks, that's good to know. It actually was an innocent
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12455117 ]
Bryan Pendleton commented on DERBY-2130:
Army, thank you for the hint about the optimizer overrides. I will investigate
this
as a possible workaround for
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12455118 ]
Bryan Pendleton commented on DERBY-2130:
Just speaking abstractly here for a minute, I want to make sure I'm following
the discussion.
As I understand
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12455141 ]
A B commented on DERBY-2130:
Everything written in Bryan's preceding comment sounds correct to me, so I
won't dwell. As for the specific questions:
I've seen
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12454963 ]
A B commented on DERBY-2130:
Just a random though that occured to me...
The query in question is written as:
prepare p1 as '
select count(distinct t1.c5)
from
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12454745 ]
A B commented on DERBY-2130:
Some items might be:
1) update the store costs to reflect current machines
o update the store costs to reflect actual machine it is
* I did not see the non-deterministic behavior that Bryan did
with 10.2 trunk--instead, I consistently see 7+ minutes. However,
I only ran the query 3 times, which is probably not enough to have
seen what Bryan saw. Also note that I only ran the DDL one time
and then I re-ran the query
just so we are working on the same case, when you run is there
also 0 rows in all the tables?
Bryan Pendleton wrote:
* I did not see the non-deterministic behavior that Bryan did
with 10.2 trunk--instead, I consistently see 7+ minutes. However,
I only ran the query 3 times, which is
Mike Matrigali wrote:
just so we are working on the same case, when you run is there
also 0 rows in all the tables?
Correct. I am just running that exact script from the JIRA,
and all it does is issue all the CREATE TABLE/INDEX/VIEW
statements and then prepare the query.
So the tables are
[
http://issues.apache.org/jira/browse/DERBY-2130?page=comments#action_12454479 ]
Bryan Pendleton commented on DERBY-2130:
One experiment I tried was to remove all the secondary
indexes. I thought that that might shrink the overall
30 matches
Mail list logo